Catalytic Hydrodeoxygenation
Catalytic Hydrodeoxygenation
Catalytic Hydrodeoxygenation
Review
Catalytic hydrodeoxygenation
Edward Furimsky∗
IMAF Group, 184 Marlborough Avenue, Ottawa, Ont., Canada K1N 0G4
Received 28 September 1999; received in revised form 6 December 1999; accepted 7 December 1999
Abstract
The importance of hydrodeoxygenation (HDO) which occurs during hydroprocessing depends on the origin of feeds. HDO
plays a minor role in the case of the conventional feeds, whereas for the feeds derived from coal, oil shale, and, particularly from
the biomass, its role can be rather crucial. The mechanism of HDO was established using a wide range of model compounds.
Complexities in the HDO kinetics have been attributed to the self-inhibiting effects of the O-containing compounds as well
as inhibiting and poisoning effects of the S- and N-containing compounds present in the feeds. This is a cause for some
uncertainties in establishing the order of the relative HDO reactivities of the O-containing compounds and/or groups of the
compounds as well as relative rates of the removal of S, O and N. Complexities arise particularly for real feeds. This is
supported by deviations from the established order such as HDS>HDO>HDN. The cases for which the overall HDN was
greater than HDO were also observed. In this case, distribution of the O- and N-containing compounds in the feed and the
type of catalyst are of a primary importance.
HDO is the main reaction which occurs during hydroprocessing of the bio-feeds. The current research activities in HDO
are predominantly in this area. Apparently, more stable catalysts are needed to make production of the commercial fuels from
the bio-feeds more attractive. © 2000 Elsevier Science B.V. All rights reserved.
Keywords: Hydrodeoxygenation; Furan rings; Phenols; Hydroprocessing catalysts; Coal-derived liquids; Bio-oils
0926-860X/00/$ – see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 6 - 8 6 0 X ( 9 9 ) 0 0 5 5 5 - 4
148 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
performance during the fuel combustion. During hy- pressure and a higher temperature are necessary for
droprocessing, such compounds may be the cause of their conversion to O-free products. On account of
a rapid catalyst deactivation. high phenols’ content, conditions employed during the
The first review on HDO was published in 1983 hydroprocessing of stabilized bio-oils may approach
[1]. Limited information on various aspects of HDO those employed during the upgrading of CDLs.
was available in the literature at that time. Therefore, During the HDO studies, conventional hy-
some conclusions had to be based on the assump- droprocessing catalysts, such as CoMo/Al2 O3 and
tions and/or on the information extrapolated from the NiMo/Al2 O3 , were used most extensively. In this
HDS and HDN results available in the literature. At review, their properties, structure and composition
the same time, considerable interest in the upgrading will receive little attention because these aspects of
of the coal-derived liquids (CDLs) was noted. This the catalysts have already been reviewed extensively
resulted in numerous studies on the HDO involving [2–5]. The most recent account of the utilization of
model O-compounds, mixtures and real feeds. In addi- hydroprocessing catalysts was given by Grange and
tion, the understanding of HDO was expanded by nu- Vanhaeren [6]. The reactions which occur simultane-
merous studies on upgrading of the biomass-derived ously with HDO, i.e., HDS, HDN and HDM, have
liquids, and to a lesser extent, also that of oil shale. been reviewed periodically as well. Comprehensive
Considering the wealth of the new information, an up- reviews of the reactions occurring during hydropro-
date of the review appears to be desirable. Thus, a cessing were published by Girgis and Gates [7] and
database of the experimental results available in the lit- Topsøe et al. [8]. Catalyst deactivation during hy-
erature can now be accessed and used to clarify several droprocessing, including the adverse effects of the
issues, i.e., effects of the catalyst type, feed composi- O-compounds, was reviewed recently by Furimsky
tion, processing conditions etc. on HDO. The issues and Massoth [9], as well as a potential reuse of the
which require additional research may be addressed. catalysts after regeneration [10]. The aspects asso-
It should be noted that HDO has also attracted atten- ciated with the selection of the catalysts to match
tion because hydroprocessing may convert waste plas- a particular feed with a reactor and hydroprocess-
tics and other O-containing wastes to usable products. ing conditions have also received attention [11]. It
Other potential applications, in which HDO can play is believed that advanced hydroprocessing catalysts,
certain role, may emerge in the future. such as those used for deep HDS and aromatics re-
The H2 consumption and severity of the opera- moval, possess also a high HDO activity. In recent
tion required for achieving high HDO conversions years, novel metal carbides and metal nitrides were
depend on the content and type of the O-compounds successfully tested for HDO and other reactions [12].
in the feed. An active catalyst must be present to However, it may take some time before these catalysts
achieve desirable HDO conversions. For CDLs, more find a commercial use.
than one stage may be required to achieve complete
HDO because of a high content of O-compounds and
aromatics. A multi-stage operation is an essential re- 2. Oxygen compounds
quirement for conversion of bio-oils to commercial
fuels. The first stage, the so-called stabilization stage, Approximate compositions of feeds differing
is performed below 573 K to remove O-compounds in origin are shown in Table 1. Structures of the
which readily undergo polymerization. In this case, O-compounds identified in the feeds to be used for
primary reactants such as methoxyphenols, biphenols the production of fuels are shown in Fig. 1. Thus,
and ethers are converted to phenols which have to the feeds include conventional liquids, cracked distil-
be removed in the second stage performed at about lates, heavy oils, fractions derived from the primary
623 K. Other O-compounds, i.e., ketones, carboxylic upgrading of heavy oils, CDLs, oil shale liquids and
acids, esters etc. may also be involved. Furans and bio-oils. The proportions of the O-groups will depend
phenols (mostly methylated) are the predominant on the origin of the feeds [1].
O-compounds in CDLs. Some of these compounds The O content of the conventional petroleum-derived
have a low HDO reactivity. Therefore, a higher H2 feeds is less than 2 wt.% [13]. The average value of
E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190 149
Table 1
Compositions of some feeds for HDO
Conventional crude Coal-derived naphtha Oil shale crude Bio-oils
Liquefied Pyrolyzed
the O content estimated by the American Petroleum actions play a minor role during the hydroprocessing
Institute was about 0.5 wt.%. The O content increases of light fractions, their importance increases during
with the boiling point of the fractions derived from catalytic upgrading of heavy residues. Naphthenic
petroleum. Thus, the O content of the asphaltenes acids were perhaps the first O-compounds identified
and resins which were separated from the distillation in petroleum. During distillation, they concentrate
residue may approach 8 wt.%. Then, while HDO re- in the gas oil and vacuum gas oil [14]. It has been
concluded that the carboxylic acids with carbon less acids were linear and were, predominantly, of the
than C8 are aliphatic. Monocyclic acids begin at C6 C14 –C20 range. Species containing carboxylic groups
and predominate above C14 . A very small amount and quinones were identified by Novotny et al. [27]
of the O in conventional liquids complicates the and Boduszynski et al. [28], respectively.
identification of the other individual O-compounds. A review of the processes used for the production of
Nevertheless, groups such as hydroxyl, carbonyl, bio-oils shows a wide range of composition, although
carboxylic, etheric and sulphoxides were found in the processes can be grouped into two general groups,
conventional fractions [1]. It is believed that at least a i.e., high pressure liquefaction and pyrolysis [29]. The
part of these compounds arose from the exposure to O content of the primary liquids from pyrolysis may
air. Thus, it is well established that such groups are approach 50% [30], whereas that from liquefaction is
formed during the autoxidation of hydrocarbons and less than 25% [29]. An extensive characterization of
thiophenes [15]. With respect to HDO, the most im- bio-oils from pyrolysis was undertaken by Maggi and
portant O-compounds are phenols and furans because Delmon [30,31]. Typical O-containing structures iden-
of their resistance to HDO. During cracking, phenols tified in these studies are shown in Fig. 2. Phenols may
are converted to arylethers, indicating their presence account for one-fourth of liquids derived from ligno-
in cracking distillates. cellulosic biomass [32]. Other types of O-compounds
HDO is among the key reactions occurring dur- include ketones, aldehydes, carboxylic acids, esters,
ing the fuel production from CDLs. Depending on alcohols and ethers [33]. Compounds containing two
the coal and the process, the O content in the feed or more O-groups, i.e., hydroxylic and etheric groups,
may approach 10 wt.%. The type of O-compounds are quite common. Examples of such structures are
depends on the liquefaction process and structure of guaiacols (GUAs), eugenol, vanillin and biphenols.
coal from which the liquids were derived as well. An Two steps might be required to achieve a complete
extensive characterization of the CDLs from the sol- conversion of such compounds to hydrocarbons. Ad-
vent refined coal (SRC) process was undertaken by ditional phenols and dihydroxyphenols are important
Gates et al. [16–21] for the purpose of studying the products of the first step, the so-called stabilizing step.
HDO reactions occurring during hydroprocessing. Us- Therefore, the HDO of such a species is an important
ing preparative liquid chromatography, these authors part of the final step.
separated nine fractions from an SRC liquid. The
O-compounds, predominantly of a phenol type, were
3. Thermochemical aspects of HDO
concentrated in a very weak acid fraction and weak
acid fraction containing 8.90 and 9.79 wt.% O, re-
Several chemical bonds have to be broken before
spectively. The following compounds were identified
the final elimination of O. Some bond strengths are
as major components: 5,6,7,8-tetrahydro-1-naphthol,
shown in Table 2 [34]. In the case of ethers and alco-
2-hydroxyphenylbenzene, 4-cyclohexylphenylphenol
hols/phenols, the bond strength of the O attached to
and an unidentified isomer of methylphenol. Other
the aromatic carbon (CAR ) is about 84 kJ/mol greater
O-compounds, i.e., furans, ethers and ketones, were
than that of the O attached to the aliphatic carbon
concentrated in a neutral-oil fraction. Hydroxy-
(CAL ). This implies that O elimination from phenols
pyridines and hydroxyindoles were found in a
and aromatic ethers will be more difficult than from
basic-fraction. Alkylphenols and alkylindanols were
alcohols and aliphatic ethers. After HYD of the aro-
predominant components of the CDL fractions from
matic ring to corresponding cycloalkane, the CAR –O
the Exxon donor solvent (EDS) process [22]. The liq-
bond is converted to CAL –O bond. This will enhance
uid remaining on the catalyst surface after hydropro-
O elimination. Then, H2 pressure may be an important
cessing of a CDL contained arylethers, xanthenes
factor influencing HDO. Some HYD equilibria were
(XA), furans and phenols [23]. Predominantly, mono-
estimated at 623 K and are shown in Fig. 3. These cor-
cyclic phenols, in addition to naphthols, indanols
relations were estimated from the following equation:
and fluorenols, were identified in shale oil by Bett
et al. [24] and Rovere et al. [25]. Afonso et al. [26] α
logKp = log − m logP
found 1.2 wt.% of carboxylic acids in a shale oil. The 1−α
E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190 151
Table 2
Bond dissociation energies (kJ/mol)
RO–R 339
RO–Ar 422
R–OH 385
Ar–OH 468
Fig. 3. Hydrogenation equilibria of model compounds.
152 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
Table 3
However, styrene was an important product during the
Reactions corresponding to Fig. 4 [36]
HDS of the S analog benzothiophene (BT) at a near
1 BF+H2 =2,3-dihydrBF atmospheric pressure of H2 [37]. It was suggested on
2 BF+2H2 =styrene+H2 O
the basis of bond strength considerations that styrene
3 2,3-dihydroBF+H2 =o-ethylphenol (OEP)
4 OEP+H2 =ethylbenzene+H2 O formation from BT may be more favorable than from
5 OEP+3H2 =ethylcyclohexanol BF [38]. Similar evaluation of dibenzofuran (DBF)
6 OEP+H2 =phenol+C2 H6 cannot be done because necessary thermodynamic
7 Ethylbenzene+H2 =toluene+CH4 data is lacking. A direct extrusion of O from DBF
8 Ethylbenzene+H2 =benzene+C2 H6
avoiding ring HYD is supported by the strength of
9 Ethylbenzene+3H2 =ethylcyclohexane
the newly formed chemical bonds compared to that
of the broken bonds. Thus, the dissociation energy of
the CAR –CAR bond connecting two benzene rings in
the subsequent HDO of the ethylphenol intermediate biphenyl and two newly formed CAR –H bonds, i.e.,
will govern the overall HDO of BF. about 493 and 468 kJ/mol, respectively, are among
Edelman et al. [36] determined the effect of tem- the strongest organic bonds. Also, direct extrusion
perature on equilibrium constants for several reactions may be favored by a high stability and symmetry of
(Table 3) which may be part of the overall HDO of BF. the final product (biphenyl). Further, the following
These correlations are shown in Fig. 4. It appears that order of the ring resonance energy was established:
direct HDO of BF yielding styrene and H2 O is ther- furan<BF<DBF [39], suggesting that HYD of the
modynamically more favorable than the ring HYD, DBF will be least favorable and that of the furan most
although styrene is never experimentally detected favorable. Then, the probability of the direct O extru-
under H2 pressure typically used in industrial units. sion will increase from furan to DBF. This approach
was used to compare relative easiness of the removal drocarbons, was part of the overall mechanism, sug-
of S, N and O from petroleum [38]. gesting that CO should be among the products [41].
Chary et al. [42] compared carbon-supported CoMo
and NiMo catalysts with the Al2 O3 -supported CoMo
4. Mechanism of HDO and NiMo catalysts at a near atmospheric pressure
of H2 during the HDO of furan. The HDO activity
Essential information for elucidation of the HDO of the former was greater than that of the Al2 O3
mechanism has been obtained during studies involv- supported catalysts. The products from the HDO of
ing model compounds. Most of the attention has been methyl-furan included isomers of pentene, pentane
paid to furans and phenols. The latter are either present and a small amount of pentadiene [43], whereas the
in the feed or are formed as intermediates during the HDO of 2,5-dimethyl-furan yielded 1-hexenes and
overall HDO of the furans. Interests in bio-oils was the 1,5-hexadiene as the major products [44]. The HDO
reason for including other model compounds in HDO of methylfuran was severely poisoned by piperidine
studies, e.g., methoxy hydroxybenzenes (GUAs), di- and to a much lesser extent also by lutidine. On ac-
hydroxybenzenes, methylarylethers, carboxylic acids count of a near atmospheric pressure of H2 used,
etc. these results reflect conditions of a limited hydrogen
HDO occurs simultaneously with competitive reac- availability, i.e., for a deactivated catalyst.
tions such as HDS, HDN, HYD and HDM, which will Thermodynamic equilibrium calculation revealed
influence the overall HDO mechanism. Therefore, that, under an H2 pressure which is typical for hy-
the experimental conditions employed, feed compo- droprocessing, the furan ring is completely hydro-
sition and type of catalyst used must be thoroughly genated to tetrahydrofuran (THF) [38]. Then, the
examined before comparing the reaction networks hydrogenated ring is an important intermediate be-
proposed by different workers. For example, it is de- fore the final O elimination can take place. Under the
sirable that the feed contains enough of a sulphur do- same conditions as applied during the HDO of furan
nating agent to prevent modification of the sulphided [40], i.e., a near atmospheric pressure of H2 , the HDO
form of the catalyst by H2 O. Difference in the HDO of THF was about three times faster than that of the
mechanism can arise by comparing studies on the furan [45]. Also, butadiene was an important product,
individual model compounds with those of mixtures particularly over the reduced CoMo/Al2 O3 catalysts.
of model compounds. The reactions occurring during As is shown in Fig. 5, the overall HDO conversion
the hydroprocessing of real feeds can be discussed in was higher on the presulphided catalyst [46]. This was
general terms only. complemented by a lower coke laydown. The temper-
ature increase from 603 to 703 K resulted in a signif-
4.1. Furans icant increase in the butadiene yield. Other products
included all the same C3 and C4 compounds which
The HDO of furan was conducted at 673 K and at were formed during the HDO of furan. Kordulis et al.
near atmospheric pressure of H2 in the presence of [47] observed that the overall HDO conversion of THF
a reduced and sulphided CoMo/Al2 O3 catalyst [39]. over the fluorinated NiMo/Al2 O3 catalyst increased
For the latter, the overall HDO conversion was more compared to the fluorine-free catalyst. However, they
than twice that observed over the reduced catalyst. did not study the effect of fluorine on the product dis-
The products included ethylene, propane, propene, tribution. Bartok et al. [48] conducted HDO of THF
n-butane, 1-butene and cis- and trans-butenes. The over Pt catalysts supported on TiO2 , SiO2 and Al2 O3
yield of n-butane decreased with time on stream, in- between 423 and 623 K. The support had a pronounced
dicating a decreased availability of the surface hydro- effect on the selectivity. Thus, on the Pt/TiO2 cata-
gen due to deactivation. It was suggested that butane lyst, the main product was butane, while on the other
and butenes arose from a partially hydrogenated ring two catalysts, decarbonylation accompanied by CO
and/or HYD of butadiene, while still being adsorbed formation was predominant. The butoxy species, still
on the surface. In addition, hydrocracking of the C–C attached to the catalyst surface, was proposed to be
bond attached to the O heteroatom, giving C3 hy- a key intermediate for product formation. In a similar
154 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
closure with increasing DBF conversion. Fig. 12 from the HDO of phenylphenol and cyclohexylphenol
shows that the mass balance closure could be ob- observed by LaVopa and Satterfield [59]. As Fig. 13
tained only by assuming that the single-ring products shows, the HDO of 2-cyclohexylphenol gave about
accounted for the difference. It can be questioned 90% single-ring products compared to less than 20%
whether single-ring products arose directly from the for the HDO of the phenylphenol. Cyclohexane ac-
parent reactant or were formed in secondary reactions counted for about 80% of the single-ring products.
of the hydrogenated intermediates. The latter route The profiles in Fig. 13 suggest that dicyclohexyl was
is preferred on the basis of the product distribution a precursor to cyclohexane formation.
4.2. Phenols
Fig. 12. Effect of space time on formation of two-ring products from HDO of DBF [60].
were similar. As expected, di-o-methylphenol was although these were only minor products. Very small
the least reactive and it underwent a greater dealkyla- quantities of these products were detected by Laurent
tion than o-methylphenol. The catalyst presulphiding and Delmon [71] and other workers, particularly at
had a pronounced effect on the product distribution, high reactant conversions.
particularly on the yield of the hydrogenated parent A study on the HDO of cresols, published by
product. tert-Butylphenol was completely dealky- Wandas et al. [72], was conducted in the pres-
lated to benzene and cyclohexane. Thus, its HDO ence of naphthalene over a CoMo/Al2 O3 catalyst at
involved that of the phenol. Benzene rings containing 633 K and 7 MPa of H2 . Conversion of cresols in
one more methyl group than the parent reactant were the presence of naphthalene was lower than that of
also detected, although in small quantities. Similarly, cresols reacting individually. Also, in the former case,
2,6-tert-butyl-4-methylphenol was completely dealky- a considerably greater variety of compounds was
lated in both 2,6-positions, whereas the 4-position formed in addition to methylcyclohexane, toluene
remained unchanged [70]. The overall mechanism for and ethylcyclopentane, which were the main prod-
the HDO of the o-substituted phenols shown in Fig. ucts during the HDO of single cresols. New prod-
15 includes two main HDO reactions, i.e., direct HDO ucts included cyclohexane, dimethylcyclohexanes
and HDO via hydrogenated phenol, occurring in par- and xylenes (mainly m-xylene in the case of o- and
allel. In the latter case, H2 O elimination may result p-cresols), as well as the O-containing intermedi-
in the formation of the intermediate methylcyclo- ates such as phenol, o-cresol arising from p-cresol,
hexene species, which will be hydrogenated rapidly. dimethyl and even trimethylphenols. Thus, in the
The formation of cyclohexene, alkylcyclohexenes case of o-, p- and m-cresols, the predominant species
and methylcyclopentanes is also shown in Fig. 15, were 2,6-dimethylphenol, 2,4-dimethylphenol and
E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190 159
whereas for the latter, various phenols accounted for the stabilized liquids can be described in terms of the
most of the converted reactant. HDO of phenols.
Guaiacol (GUA) and substituted GUAs (Fig. 20a)
4.4. Bio-oil compounds have attracted much attention because of their rela-
tively high content in bio-oils and low stability. The
Fig. 20a was added to assist the reader with the un- mechanism shown in Fig. 21 indicates coke formation
common structures of O-compounds used frequently [90]. Coke was formed from both GUA and the pri-
to study the HDO mechanism of bio-oils. They in- mary product catechol (CAT). This is clearly shown
clude dihydroxyphenols, alkyl-arylethers, ketones, in Fig. 22. Thus, a mass balance closure could not be
carboxylic acids, esters and alcohols. In most stud- achieved. The reaction network proposed by Delmon
ies, conventional hydroprocessing catalysts have been et al. [84,85] and Bredenberg et al. [86–88] over a
used. Some of these compounds readily polymerize CoMo/Al2 O3 catalyst considers the hydrogenolysis
to coke-like products, even during their distillation. of the methoxy group to CAT and methane as the first
Then, a stabilization step, usually performed between stage, followed by the elimination of one OH group
473 and 573 K, may be required, particularly if a pro- from CAT in the second stage to produce phenol.
longed storage of such liquids is anticipated [83–93]. The coke arises from the interaction of GUA with the
Most of the studies involving these compounds fo- ␥-Al2 O3 support rather than with the active metals
cussed on the stabilizing step only, i.e., phenols were [85]. CAT gave a similar amount of coke as the latter
the major O-containing products. Then, the HDO of when treated under the same conditions separately.
Fig. 23. Yield of phenol as a function of yield of catechol, CoMo on Al2 O3 (䊉) and carbon (4) [84].
164 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
Fig. 26. Plot of (Pox /PR,O )2 vs. W/F for HDO of BF [36].
where Pox =PR +P1 +P2 +P3 . Experimental data in the ference between the experimental conditions of these
form −ln(Pox /PR,O ) versus W/F showed that the ob- studies is emphasized. First of all, the reactant con-
served HDO does not follow first-order kinetics above centration of the former was much higher than that
623 K and high space times. Thus, an increase in khdo used by Lee and Ollis [50], indicating a more exten-
with decreasing Pox suggests a self-inhibiting effect sive self-inhibition by the reactant. Also, the ratio of
of the oxygenated compounds. A zero-order model BF to sulphur agent was about 20 times greater in the
examined by Edelman et al. [36] was also rejected. study conducted by Edelman et al. [36]. Then, H2 S in
Subsequently, these authors examined the following excess of that required to maintain catalyst in a sul-
model: phided form inhibited the HDO reactions [48].
−d(Pox /PR,O ) k(Pox /PR,O )
= Table 6
d(W/F ) {1 + K(Pox /PR,O )}2 Rate constants for hydrogenation and HDO of BF
After a simplification, this equation was integrated to Temperature (K) khyd ×107 (l/s g cat) khdo ×107 (l/s g cat)
yield Edelman et al. [36]a
573 10.2 0.5
Pox 2 W
= 1 − 2khdo 623 28.8 5.0
PR,O F 673 46.7 14.0
Lee and Ollis [50]b
A plot of (Pox /PR,O )2 versus W/F, shown in Fig. 26, 573 – 3.0
appears linear for the (−1)-order model. An apparent 598 – 6.5
HDO activation energy estimated from these values ak
hyd : pseudo-first-order; khdo : (−1)-order; concentration of
was 106 kJ/mol. BF: 0.6 mol/l; NiMo/Al2 O3 ; 0.3 mol of dimethyl sulphide.
The rate constants estimated by Edelman et al. [36] bk
hdo : pseudo-first-order; concentration of BF: 0.15 mol/l;
and Lee and Ollis [50] are compared in Table 6. Dif- CoMo/Al2 O3 ; 0.0075 mol of mercaptan.
E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190 167
Table 7 Table 9
Rate constants for HDO of DBF Pseudo-first-order rate constants for HDO of DBF (l/g cat s×106 )
[75]
Temperature (K) khdo ×106
Single compound
LaVopa and Satterfield [59]a
Single-ring products 16.0
623 4.8
Biphenyl 0.6
633 6.5
Phenylcyclohexyl 1.9
648 8.3
663 10.7 In mixture A (single-ring products)
with 5,6,7,8-tetrahydronaphthol 5.8
Girgis and Gates [75]b
without 5,6,7,8-tetrahydronaphthol 13.0
623 18.1
a In neutral oil
7 MPa; 0.245 mol/l; 13.9 kPa H2 S.
b 17 MPa; 0.0004 mol/l; 0.0001 mol/l 3,7-dimethylDBT. DBF disappearance 7.4
of DBF was poisoned by NH3 [98] and quinoline actor [18,21,50,51,60,62–64,80] and in a batch reactor
[47]. [68,71,73,100]. Sulphided commercial CoMo/Al2 O3
The difference in experimental conditions used in and NiMo/Al2 O3 catalysts were the most frequently
various HDO studies prevents direct comparison of used catalysts.
the rate constants. To overcome this problem, Girgis The batch reactor study conducted by Gevert et al.
and Gates [75] expressed rate constants relative to that [68] focussed on the HDO of 2-, 4- and 2,6-substituted
of the most rapid reaction in the overall HDO mech- phenols. The simplified reaction network used is
anism of DBF to compare their results with those ob- shown in Fig. 29. The following equations to re-
tained by Krishnamurthy et al. [57]. The results in late mole fractions of the phenols (XA ), aromatics
Table 8 show a good agreement between the two stud- (XB ) and cyclohexane+cylohexene (XC ), with the
ies. The comparison became a challenge when kinetic pseudo-first-order rate constants, were derived:
data was obtained for a single compound in a mixture
k1
with other heterocyclics and aromatics. Gates et al. XB = (1 − XA )
[7,60,99] studied the HDO of DBF in the mixture con- k1 + k2
taining pyrene, phenanthrene, fluoranthene and DBT and
with and without 5,6,7,8-tetrahydro-1-naphthol as well
k2
as that in a neutral oil fraction which was isolated from XC = XB
a CDL. The obtained pseudo-first-order rate constants k1
are compared in Table 9 together with that obtained The pseudo-first-order rate constants were calculated
for a single DBF [75]. from the following equation:
dXB
5.2. Phenols = k1 WXA
df (t/V )
Kinetic studies on the HDO of single phenols or
mixtures of phenols were conducted in a trickle bed re-
Table 8
Relative values of the pseudo-first-order rate constants
Reaction Reference
[75] [57]
Table 10
Pseudo-first-order rate constants for HDO methyl phenols over
sulphided CoMo/Al2 O3 at 573 K
Phenol+Poison k1 (l/g cat s×105 ) k2 (l/g cat s×105 )
Ref. [68]
4-MP 4.0 0.9
4-MPa 6.5 1.3
2-MP 2.1 0.3
2,4-diMP 2.5 0.3
2,6-diMP 0.5 0.2
2,4,6-triMP 0.8 0.1
4-MP+NH3 (8.5)b 0.8 0.1
4-MP+NH3 (43) 0.2 0.03
4-MP+H2 S (36) 0.5 0.8
4-MP+H2 S (72) 0.2 0.6
Ref. [79]
Phenolc 1.1
(Ref. [50]; continuous system)
2-EtPc 1.1
3-MP 1.6
aInitial concentration 70 mmol/l, all other runs 142 mmol/l.
bNumbers in brackets indicate concentration of poison in
mmol/l.
c Extrapolated from Arrhenius plots.
The initial slope of the curve for conversion versus Fig. 30. Arrhenius plots for k1 (䊊), k2 (4), 3,5-dimethylphenol
f(t/V) allowed calculation of k1 . In this equation, V is (䊊,4) and 2,6-dimethylphenol (䊉,䉱) [100].
the volume of the feed at time t. The k2 could then
be obtained in conjunction with the above equations.
The obtained rate constants are summarized in Table reactor system. The single compound experiments
10 including constants obtained in the presence of showed that p-methylphenol was much more reac-
NH3 and H2 S. The poisoning effect on the overall tive than o-ethylphenol, i.e., less than 13% of the
HDO is quite evident. The rate constants increased former’s conversion. The HDO of p-methylphenol
with decreasing concentration of substrate, indicating was inhibited by o-ethylphenol. As expected, NH3
a self-inhibiting effect of the latter on its HDO. In severely poisoned the HDO reactions, whereas H2 O
the subsequent study, Gevert et al. [100] conducted a had little effect. H2 S had a much more pronounced
detailed comparison of the 3,5- and 2,6-dimethyl phe- effect on the hydrogenolysis reactions than on the
nols. For the former, k1 was about 10 times greater HYD reactions. The effect of these agents on HDO
than that for 2,6-dimethylphenol, whereas for k2 , over a CoMo/Al2 O3 catalyst differed from that over
the difference was much less evident. The Arrhenius an NiMo/Al2 O3 catalyst (Table 11), e.g., for the
plots in Fig. 30 show that the activation energies for latter, the kMCH /kTOL was significantly greater than
hydrogenolysis were greater than for the HYD of that for the CoMo/Al2 O3 catalyst. Laurent and Del-
ring. For 2,6-dimethylphenol, HYD increased rela- mon [101] expanded their study to include high H2 O
tive to hydrogenolysis with increasing temperature, pressure conditions with the aim of simulating the
whereas an opposite trend was observed for the HDO occurring during high pressure liquefaction of
3,5-dimethylphenol. biomass. Thus, the H2 O and H2 S vapor pressure was
Laurent and Delmon [71] conducted a detailed 2.5 and 0.1 MPa, respectively, of the total pressure
study on the effect of o-ethylphenol, NH3 , H2 S and of 7 MPa. It should be noted that the H2 O saturation
H2 O on the HDO of p-methylphenol in a batch pressure under the conditions used by these authors
170 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
Table 11
catalyst. They observed approximately first-order
Pseudo-first-order rate constants for HDO of 4-methylphenol in
the presence of o-ethylphenol, H2 O, NH3 and H2 S [71]a dependence on H2 pressure and pseudo-first-order
behavior of the HDO conversion of cresols at a con-
Agent Rate constant×105
stant H2 pressure. The rate constants were determined
CoMo NiMo from the plots, such as ln(1−Xc ) versus PH2 and
kMCH kTOL kMCH kTOL ln(1−Xc ) versus W/F, respectively, where Xc is the
cresol conversion and the others are the commonly
o-Ethylphenol (mol/l)
known parameters. The rate constants estimated by
0 2.5 2.9 13.6 0.7
0.145 1.3 1.2 7.2 0.4 these authors were about two orders of magnitude
0.434 0.8 0.7 3.6 0.2 smaller than those estimated by Gevert et al. [68]
H2 O (mol/l)
and Laurent and Delmon [71]. Apparently, the cat-
0 0.9 1.0 4.2 0.2 alyst in the study of Odebunmi and Ollis [61] was
0.65 1.1 0.8 4.3 0.2 already deactivated before the measurements were
1.96 1.1 0.7 3.2 0.2 taken in addition to different experimental systems
NH3 (mmol/l) used. In subsequent studies, these authors estimated
0 1.9 2.0 6.2 0.6 the pseudo-first-order rate constant at 573 K for the
5 1.6 1.4 4.3 0.3 HDO of m-cresols [64] and o-ethylphenol [50] to be
9 1.3 1.0 3.2 0.2
in the range of those shown in Tables 10 and 11,
49 0.6 0.3 1.1 0.1
i.e., 1.6×10−5 and 1.1×10−5 l/g cat s, respectively.
H2 S (mmol/l) Temperature can influence relative reactivities of
0 0.7 3.7 5.3 0.5
17 0.8 2.0 5.3 0.3
cresols. This is supported by the results in Fig. 31
49 0.9 1.0 4.2 0.2 [62] which show that, for o- and p-cresols, the differ-
98 1.0 0.7 3.9 0.2
H2 O (25 bar) 0.85 0.05
H2 O+H2 S (25 bar+1.1 bar) 1.37 0.06
a Conditions: batch strirred reactor, 613 K, 7 MPa of H , do-
2
decane solvent; 0.327 mol/l.
Fig. 32. Effect of inverse space velocity on HDO conversion of model compounds [20].
ence in reactivity decreases with increasing temper- dicates inhibition of HDO by other compounds. The
ature. Activation energies for the HDO of o-, m- and rate constants in Table 12 were obtained at 623 K,
p-cresols estimated from these results were 96, 113 compared to those in Tables 10 and 11 which were
and 156 kJ/mol, respectively, compared to 125 kJ/mol obtained at 573 K. Besides the different experimen-
for the phenol estimated in a batch reactor [79].
Gates et al. [7,20,21] conducted studies on the Table 12
HDO of phenols in mixtures isolated from a CDL Pseudo-first-order rate constants for disappearance in mixtures (l/g
over a sulphided NiMo/Al2 O3 catalyst. The very cat s×104 )a
weak acid and weak acid fractions contained only [21] [20]
low concentrations of sulphur and nitrogen and high
Pure compound in cyclohexane
concentrations of the O-containing compounds, i.e.,
2-Hydroxyphenylbenzene 1.5
8.90 and 9.79 wt.%, respectively [16]. These fractions Weak acid fraction in cyclohexane
were used as 0.25 wt.% solutions in cyclohexane. In 2-Hydroxyphenylbenzene 0.83 1.8
one study, the product analysis was performed using Methylphenylphenol 1.52 2.0
IR and GC-MS techniques [21], whereas in the other 4-Cyclohexylphenol 4.46 6.2
5,6,7,8-Tetrahydro-1-naphthol 1.91 4.2
study, the GC-MS technique was used to identify
Methyltetrahydronaphthol – 4.5
compounds and FID chromatography to determine Dimethyldihydroxyindan – 4.6
their concentrations [20]. The results in Fig. 32 and 1-Naphthol – 5.4
Table 12 show that the method of analysis influences Phenolics in weak acids 1.56
the values. Comparison of the rate constants for pure a Conditions: flow reactor; sulphided NiMo/Al2 O3 ; cyclohex-
2-hydroxyphenylbenzene and that in the mixture in- ane solvent; 623 K; 12 MPa.
172 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
Table 13
stant for hydrogenolysis of diphenylether at 623 K
Activation energies of reactions in Fig. 18 [73]
and about 14 MPa pressure of H2 for CoMo/Al2 O3
Reaction Activation energy (kJ/mol) and NiMo/Al2 O3 catalysts, as well as the Mo/Al2 O3
1 139±32 catalysts promoted by the group VIII metals. For the
2 100±23 three different CoMo/Al2 O3 catalysts, the rate con-
3 44±20 stant varied between 2×10−4 and 4×10−4 l/g cat s,
4 132±38
whereas that for an NiMo/Al2 O3 catalyst was about
5 77±23
1×10−4 l/g cat s. Apparently, in this study [82], phe-
nol was the primary product but was rapidly converted
to hydrocarbons. Thus, the rate constants approach
tal systems used, i.e., batch versus continuous, other that for an overall HDO. Artok et al. [80] observed
parameters may be responsible for the difference in that, besides different product distribution (indicated
values. Of particular importance is the concentration above), the HDO of dinaphthylether was much faster
of reactant, which in the case of the Gevert et al. [68] than that of the diphenylether. They suggested that
and Laurent and Delmon [101], was more than an the lower resonance stability of the phenoxy radicals
order of magnitude greater than that used by Gates compared to that of the naphthoxy radicals was at
et al. [7,20,21], indicating a much lower degree of least partly responsible for the lower reactivity of the
self-inhibition. A higher H2 pressure used by the diphenylether. Kirby et al. [70] compared XA with
latter authors may also contribute to the difference. dinaphthylether and found the former to be much less
Kinetics of the HDO of naphthol were investi- reactive. However, the study was conducted in the
gated by Li et al. [73] over a sulphided NiMo/Al2 O3 presence of a Co–Mo organometallic precursor rather
catalyst. The pseudo-first-order rate constants esti- than a typical hydroprocessing catalyst.
mated at 427 K for all steps are shown in Fig. 18.
It is evident that, at this temperature, direct HDO of
naphthol is slower than ring HYD. However, above 5.4. Bio-oil compounds
550 K, direct HDO exceeded HYD. This is supported
by the activation energies shown in Table 13. The The detailed study on the kinetics of bio-oil related
pseudo-first-order rate constants for the disappearance model compounds was published by Laurent and Del-
of the naphthol and 5,6,7,8-tetrahydro-1-naphthol mon [83,91]. The work was conducted in a batch re-
shown in Table 12 were estimated for a diluted acidic actor from 523 to 573 K and an H2 pressure of 7 MPa
fraction derived from a CDL. Based on this data, it in the presence of commercial CoMo/Al2 O3 and
is suggested that the HDO reactivity of naphthol is NiMo/Al2 O3 catalysts kept in a sulphided form. The
similar to that of m- and p-substituted phenols. conversion data was fitted to the following first-order
kinetic equation:
5.3. Ethers −lnXi = kWt
The pseudo-first-order rate constants (l/s g cat) for where Xi is the ratio of the concentration of the re-
the disappearance of ANI, estimated by Hurff and actant in the sample (Ci ) to the concentration of the
Klein [102] over a sulphided CoMo/Al2 O3 catalyst at reactant in the initial sample (C0 ), k is the pseudo
523, 548 and 598 K, were 0.0763×10−3 , 0.603×10−3 first-order rate constant (min−1 g cat−1 ), W is the
and 2.78×10−3 , respectively. However, phenol was weight of catalyst and t is the time. Pseudo-first-order
the major product. Then, the HDO of phenol will logarithmic plots for 4MA, DES and GUA are shown
govern the overall HDO of ANI. In this temperature in Fig. 33. In most cases, the experimental data did not
range, a value for the activation energy of 124 kJ/mol follow pseudo first-order over the whole conversion
was obtained. Dibenzylether and benzodioxan (sul- range, i.e., only the first points were used in the de-
phided NiMo/Al2 O3 , 648 K and 6.9 MPa) were termination of the rate constants. Deviation from the
much more reactive than m-ethylphenol [51]. Shabtai first-order kinetics occurred at high conversions. Fast
et al. [82] determined the pseudo-first-order rate con- and slow conversion ranges were observed for GUA.
E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190 173
Table 15
Pseudo-first-order rate constants for disappearance of 4-MA, DES and GUA and % DES decarboxylation and GUA phenol/CAT and coke
on catalyst after reaction [91]
Catalyst Rate constant (cm3 /min g cat) Coke (wt.%)
k4MA kDES % Decarboxylation kGUA Phenol/CAT
in a continuous system in the presence of a typi- latter factor is important when catalyst deactivation af-
cal hydrocracking catalyst, i.e., NiMo/SiO2 ·Al2 O3 , fects the reaction. Other parameters include H2 pres-
observed ANI to be more reactive. In this case, the sure, type and form of catalyst, experimental system
reactants (45 mol%) were mixed with benzene. It is etc. In the case of multi-reactant systems, estimates of
assumed that, in the case of GUA, the higher acidity activation energies for HDO will be affected by HDS
of the support compared to that of ␥-Al2 O3 caused and HDN. Nevertheless, in some cases, an agreement
rapid catalyst deactivation. Also, a rather different among several workers is quite remarkable. Activa-
presulphiding procedure was used by these authors. tion energies may indicate change in relative reactivi-
Nevertheless, in spite of the differences, the product ties with the temperature change. For example, results
distribution was similar in both studies. These two obtained by Odebunmi and Ollis [62] suggest that the
studies may be used to illustrate the effect of different reactivity of p-cresol will increase relative to that of o-
experimental conditions on the final results, although and m-cresol with increasing temperature. However,
with respect to HDO, they have only a limited value as Fig. 31 shows, in the temperature region which is
because only the reactant disappearance was followed, typical of hydroprocessing, their relative reactivities
i.e., phenols were the main products. will not change. The values estimated by Li et al. [73]
The activation energies reported in several studies represent another example of how activation energy
are summarized in Table 17, as well as in Table 18, for can be used to predict the change of the overall HDO
bio-oil related compounds. Differences for the same mechanism (Fig. 18) with the change of temperature.
reaction from different workers are attributed to exper-
imental conditions which varied from study to study.
For example, for single-compound estimates, the same 6. HDO reactivities of O-containing compounds
reactant concentration was not always used, indicating
a different extent of self-inhibition. Also, temperature Relative HDO reactivities of the O-containing com-
regions and the time on stream when the estimates pounds were discussed by Landau [104] and Afonso
were made varied. As the results in Table 18 show, the et al. [26], who pointed out complexities in estab-
lishing a true order. For the purpose of this discus-
sion, reactivity is defined as the overall conversion of
Table 16 an O-compound to a hydrocarbon rather than to an
Acidity of fresh catalysts and their coke content after reaction [91]
O-containing intermediate. Grange et al. [105] used
Acidity (eq NH3 /g) Coke (wt.%) the iso-reactivity, i.e., the temperature at which a sig-
␥-Al2 O3 359 10.4 nificant identical value of conversion (to hydrocar-
CoMo/Al2 O3 522 9.0 bons) can be attained in the presence of a commercial
CoMo/Al2 O3 /K 456 10.0 hydroprocessing CoMo/Al2 O3 catalyst. The reactiv-
CoMoSiO2 113 2.7 ity trends established from these values are shown in
CoMoC 111 –
CoMoS n.d. 2.9
Table 19. They pointed out that aliphatic ethers and al-
cohols are even more reactive than ketones. Böhringer
E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190 175
Table 17
Summary of activation energiesa
Reaction Catalyst Activation energy (kJ/mol) Reference
and Schultz [106] introduced the T50HC parameter, de- in which the rate constants are expressed differ from
fined as the temperature at which 50% conversion to study to study. In some cases, the units do not take
O-free products was achieved. into consideration the reactant concentration (e.g.: l/g
The overall HDO conversions and the rate constants cat t and/or g/g cat t), although a self-inhibiting effect
determined during the kinetic measurements are suit- of some reactants on their HDO is well documented.
able parameters for determining relative HDO reac- Most of the results on the HDO of the bio-oils are
tivities. However, differences in experimental condi- based on reactant disappearance rather than complete
tions prevent a direct comparison of the data obtained HDO. Thus, the stabilizing step is the primary objec-
by different workers. The results in Tables 17 and 18 tive, assuming that complete HDO will be achieved
show that activation energies for the HDO of indi- during the next step. The importance of catalyst struc-
vidual reactants vary, suggesting that the relative re- ture on the relative HDO reactivities was recognized.
activities will change with temperature. Also, units Unless stated otherwise, the following general discus-
176 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
Table 18
Summary of activation energies for bio-oil compounds
Reaction Catalyst Activation energy (kJ/mol) Reference
sion assumes HDO in the presence of sulphided con- one may only speculate that 2,5-dialkyl furan will still
ventional CoMo/Al2 O3 and/or NiMo/Al2 O3 catalysts be more reactive than the unsubstituted BF. Unsub-
at temperatures, H2 pressures and other conditions, stituted furan can be excluded from these considera-
which are typical of hydroprocessing operations. It is tions. Thus, it is unlikely that furan will be present in
noted that the NiMo/Al2 O3 catalyst was found to be feeds to be hydroprocessed because of its high volatil-
more acidic than the CoMo/Al2 O3 catalyst [107], i.e., ity (bp≈304 K). Uncertainties exist regarding the re-
the former has a higher cracking activity. Therefore, activity of BF, as indicated by the different values of
the type of catalyst used was always identified with the rate constants in Table 6. In this case, the differ-
the kinetic data in the preceding section. ent reactant concentrations used by different workers
Based on thermochemical considerations, the reac- are the most probable contributor. Thus, Lee and Ol-
tivity of the unsubstituted furanic rings will decrease lis [50] showed that, at 598 K, the rate constant de-
in the following order: furan>BF>DBF. The experi- creased by half when the BF concentration increased
mental data obtained under similar conditions, which from 0.15 to 0.25 mol/l. Yet, in the study of Edel-
would confirm the same order, is lacking. Further, man et al. [36], the reactant concentration was about
alkyl substitution of the furanic ring will change its 0.60 mol/l. In addition, the H2 S concentration could
reactivity. For example, in the case of furan, substitu- approach 0.3 mol/l. Apparently, the sulphur concen-
tion in 2,5-positions will decrease the reactivity due to tration in feeds exceeding an optimal level is a cause
steric effects. However, experimental data showing the of the inhibition of HDO sites by H2 S [53].
extent of the activity decrease is not available. Thus, In the study published by Böhringer and Schultz
[106], the reactivity of unsubstituted BF and DBF
was determined under the same conditions. They
Table 19 observed that the temperature at which 50% HDO
Relative reactivities of O-compounds and/or groups [105] was achieved was about 55 K higher for DBF than
Temperature of Activation for BF. This was supported by the result published
iso-reactivity (K) energy (kJ/mol) by Dolce et al. [108]. Girgis and Gates [60] showed
Ketone 476 12 that the ratio of the reactivity of DBT/DBF estimated
Carboxylic 556 26 from the overall conversion in a mixture with aro-
Methoxyphenol 574 27 matics was about 12. However, when N-compounds
4-Methylphenol 613 34 and 5,6,7,8-tetrahydronaphthol were added, the ratio
2-Ethylphenol 640 36
decreased to 3.6. In a neutral oil, this ratio was about
DBF 690 34
8 [97]. Using empirical rate constants for pure
E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190 177
compounds obtained by Lee and Ollis [53], a DBT/BF (Table 11), determined by Laurent and Delmon [71],
value of 8.9 was obtained compared to about 3.0 for is in the same range as that in Table 10. In an-
equimolar concentrations (0.15 mol/l) of the com- other batch reactor study [65], a little difference be-
pounds. The value of 12 obtained for DBT/DBF by tween m- and p-methylphenol was observed, whereas
Girgis and Gates [60] confirms a lower reactivity of o-methylphenol was the least reactive. However, the
DBF compared to BF. However, the reactant con- concentration of the reactants in hexadecane ap-
centrations and compositions of the mixtures were proached 60 mol%. Also, naphthalene was present
different. The DBT/BF and DBT/DBF ratio esti- and could interfere with the HDO. An important
mated from conversions observed by Rollman [61] batch study on the HDO of phenols was conducted by
for a mixture of compounds was 3.6 and 9.0, respec- Weigold [67] over a sulphided CoMo/Al2 O3 catalyst.
tively, indicating a lower reactivity of DBF than BF. The results of this study (Table 20) are in qualitative
Thus, all evidence suggests that, over conventional agreement with other studies, although they were ob-
hydroprocessing catalysts, the reactivity of the unsub- tained during the HDO of pure phenols. Thus, in this
stituted BF is greater than that of DBF when single case, a large self-inhibiting effect on HDO was present
compounds are considered. without any doubts. The usual order of reactivity, i.e.,
For a novel catalyst, such as Mo2 N, Abe and Bell m-methylphenol>p-methylphenol>o-methylphenol,
[55] obtained a DBT/BF ratio of 0.9, in agreement was established by Odebunmi and Ollis [62] in a
with the results published by Ramanathan and Oyama continuous system. A low reactivity of the phenol
[52]. However, the latter authors reported that, in the compared to the substituted phenols is indicated
case of VN, the DBT/BF ratio was less than 0.1, indi- by Weigold [67]. The results in Table 20 indicate
cating a superior HDO activity of the VN catalyst in a similar reactivity of the former as that of some
comparison with the commercial NiMo/Al2 O3 cata- o-substituted phenols, but much lower than that of
lyst. In the case of a V–Mo–O–N catalyst, the ratio of o-cresol. Brendenberg et al. [86] observed that, at
0.8 was observed [109]. Therefore, the type and form 598 K over an NiMo/Al2 O3 .SiO2 , the overall HDO
of the catalyst has to be considered when comparing of the phenol was about 17% compared to about
HDO reactivities of the O-compounds. It is obvious 26% for the disappearance of o-cresol. However, in
that a set of experiments still has to be designed and the latter case, phenol accounted for about 50% of
completed before the HDO reactivity of BF can be di- the product. The results in Table 10 show that the
rectly compared with that of DBF. The situation will pseudo-first-order rate constant for the HDO of the
become more complex when alkyl substituted BF and phenol is less than that of o-cresol. In the former
DBF will be compared. Nevertheless, the addition of case, the value was extrapolated from a low temper-
an aromatic ring to a furanic ring will decrease the ature region. It is noted that solid experimental data
reactivity of the compound, suggesting that a true re-
activity order will be similar to that established on the
basis of thermochemical considerations. The ring sub- Table 20
Yield of aromatic hydrocarbons from phenols (sulphided
stituted compounds of BF and DBF have not yet been
CoMo/Al2 O3 , 573 K) [67]
studied. If the observation made by Gates and Topsøe
[110] for sulphur analogs such as DBT applies also Substrate Product Yield (wt.%)
for DBF, the alkylation of the latter, particularly in Phenol Benzene 8
4- and 4,6-positions, would significantly diminish its o-Cresol Toluene 19
HDO reactivity. m-Cresol Toluene 48
p-Cresol Toluene 23
The database on the HDO reactivity of phenols is 3,4-Dimethylphenol o-Xylene 45
much more extensive than that on the furanic rings. 3,5-Dimethylphenol m-Xylene 31
The results in Table 10 [68] indicate the following or- 2,3-Dimethylphenol o-Xylene 13
der of reactivities of the methyl substituted phenols: p- 2,4-Dimethylphenol m-Xylene 7
methylphenol>o-methylphenol>2,4-dimethylphenol> 2,5-Dimethylphenol p-Xylene 10
2,3,5-Trimethylphenol 1,2,4-Trimethylbenzene 17
2,6 - dimethylphenol ∼ 2, 4, 6 - trimethylphenol. The o-Ethylphenol Ethylbenzene <1
pseudo-first-order rate constant for p-methylphenol
178 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
Table 21
complex mixtures. It is obvious that, in the same order,
Product distribution (mol%) from reaction of GUA [86]
the amount of H2 required to remove an O-containing
group will be the largest for carboxylic acids, as- Temperature (K) 548 598
suming its HYD to methyl group. This suggests that Conversion (%)
factors such as the catalyst activity, hydrogen avail- Methylcyclohexane <0.1 1.1
ability, interaction with the catalyst surface etc. will Cyclohexane <0.1 0.3
Cyclohexene <0.1 0.5
be most important for carboxylic groups. In other Benzene <0.1 0.5
words, the HDO reactivity of the carboxylic acids will Toluene <0.1 0.6
be affected by the catalyst type and its deactivation to Phenol 30.9 53.5
a greater extent than that of the more reactive groups. Anisol 1.0 2.1
This is supported by the observation made by Lau- o-Cresol 3.2 9.4
m-Cresol 2.4 7.3
rent and Delmon [88] who reported that, under the p-Cresol 0.8 1.6
same conditions, most of the DES was converted to Pyrocatechol 61.4 8.7
O-compounds (monoester and acid), whereas most of
the EDEC and DEC were converted to hydrocarbons,
although the disappearance of the diester was much
faster. Based on the observation made by Ledoux several phenols in the investigated mixture. Based on
and Djellouli [107], the NiMo/Al2 O3 catalyst is more the kinetic analysis published by Laurent and Delmon
active for cracking than the CoMo/Al2 O3 catalyst, [83,90], the following tentative order of the reactiv-
suggesting that the former will be more active for ity, defined as disappearance of the model compounds,
decarboxylation. However, the difference in activity can be established: 4MADES≥GUAANI. Based
may not be large enough to change the relative or- on the results in Table 14, the reactivity of GUA and
der of reactivities. There is some uncertainty in the DES is considered to be similar. The rate of disap-
above order about where to place arylethers because pearance of substituted GUAs (e.g., 4-methylGUA,
of their lesser sensitivity to the availability of H2 for euginol and vanillin) was compared by Petrocelli and
their overall HDO. Thus, at 673 K and 12.5 MPa of Klein [81]. The differences in the pseudo-first-order
H2 , phenols and carboxylic acids exhibited a similar rate constants for disappearance of these compounds
reactivity [26]. Phenols are intermediates during the indicate a similar reactivity for conversion to pheno-
HDO of arylethers. It is believed that when H2 avail- lic compounds and coke. The relative reactivity of the
ability becomes a factor, e.g., at a certain level of GUA and ANI is based on the results obtained by Hurf
catalyst deactivation, the overall HDO of carboxylic and Klein [102] rather than that of Bedenberg et al.
acids may be slower than that of the phenols. [86], because in the latter case, a typical hydrocracking
Studies on the HDO of bio-oils related compounds catalyst was used. In the case of the DES and 4MA,
focussed mainly on the stabilization stage, i.e., only hydrocarbon products accounted for about 50 and al-
the disappearance of the reactant was followed. As most 100%, respectively. The pseudo-first-order rate
the results in Table 21 show [86], phenols accounted constants for the disappearance of EDEC and DEC
for most of the disappeared GUA, in agreement with at 553 K over an NiMo/Al2 O3 catalyst were in the
the results published by Laurent and Delmon [83]. range of that for DES (Table 14), i.e., 9.1×10−3 and
Then, second stage treatment is required to complete 4.6×10−3 (min−1 g cat), respectively. However, the
the HDO. Little attention was paid to the HDO of the yield of hydrocarbons was much greater than that for
stabilized liquids. However, the established database DES, indicating a larger overall HDO of EDEC and
on the HDO of phenols may be a basis for deducing DEC than that of DES.
information on the overall HDO reactivity of stabi- The available information on bio-oils related reac-
lized liquids, particularly if the phenols distribution tants allows some speculation on the overall HDO re-
in these liquids is known. Thus, while studying the activities, i.e., assuming that the experiment would be
HDO of a mixture of phenols, Li et al. [21] suggested performed in one stage to achieve a near complete
that the overall HDO may be followed using one rate HDO. In such a case, the overall HDO reactivities of
constant obtained as an average of the constants for EDEC and DEC are greater than that of DES because
180 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
of the greater conversion to hydrocarbons under the cessing of a coal-derived feed published by Yoshimura
same conditions. The overall HDO reactivity of the et al. [113]. However, an excess of sulphur may have
GUA is lower than that of DES because, in the case of an adverse effect on the overall HDO [53].
the former, phenols are the predominant primary prod- Interactions of the furanic rings with the catalyst
ucts. In the former case, the HDO of hydroxyphenol surface are being frequently explained in relation to
and phenol will determine the overall HDO. Laurent similar interactions of thiophenic rings. In fact, the
et al. [82] showed that the conversion of the former generally accepted definition of the active site is based
to phenol is much faster than the conversion of phe- on evidence gathered during the HDS of the latter.
nol to hydrocarbons, suggesting that the latter reaction Also, a mutual inhibition of HDS by O-compounds
will determine the overall HDO, similar to the case of and that of HDO by S-compounds, observed by Ode-
ANI [102]. Then, with respect to overall HDO, GUAs bunmi and Ollis [63], indicates a competitive adsorp-
will be less reactive than o-phenols. A direct compar- tion at the same sites. In other words, the form and the
ison of the HDO of furanic rings with GUAs would geometry of the HDS sites and HDO sites are similar.
be required to determine the relative overall HDO re- This was also supported by the results of Yamamoto
activities. It is noted that this discussion is based on et al. [114], who observed that catalyst deactivation by
data obtained below 573 K. It is established that the coke had the same adverse effect on HDS and HDO.
HDO of phenols gains in importance with increasing It is believed that, when a hydroaromatic hydrogen is
temperature. This effect may change and/or diminish involved, this similarity will remain unchanged dur-
the relative HDO reactivities of the reactants. Other ing both HDS and HDO. Laurent and Delmon [71]
experimental parameters, e.g., contact time, H2 avail- speculated that the active site in C–O hydrogenolysis
ability etc., may be equally important. could be based on an ensemble of coordinatively un-
saturated Mo atoms and the HYD sites on one triply
unsaturated Mo atom. The H2 S concentration influ-
7. Interaction of O-compounds with catalyst ences the balance between these two types of sites.
surface during HDO This speculation was in accordance with interpreta-
tions of the adsorption of dienes and thiophenic rings
The coordinatively unsaturated sites (CUS), or sul- on MoS2 , published by Kasztelan et al. [115] and
phur anion vacancies, which are located at the edges Okamoto et al. [116], respectively. Girgis and Gates
of MoS2 slabs supported on ␥-Al2 O3 , are believed to [60] discussed the interaction of DBF in relation to
be the sites for catalytic reactions during hydropro- that of DBT. They proposed that, in the latter case, the
cessing. These sites can adsorb molecules with un- interaction occurs through the bonding of C1 –C2 bond
paired electrons such as NO, NH3 , and pyridine, i.e., at an anion vacancy. This makes electron distribution
they have a Lewis acid character. The vacancies can around the sulphur more deficient. This promotes ring
consist of a significant fraction of the edge sulphur interaction with a surface sulphide anion. For furanic
atoms. Double and even multiple vacancy centers can rings, similar electron distribution changes occur to
be present. The presence of Co and Ni does not af- a lesser extent because oxygen is much less polariz-
fect the basic slab size of the MoS2 . The Co or Ni able than sulphur. This will weaken the adsorption of
does not appreciably increase the number of vacan- furanic rings on the catalyst surface. Thus, a higher
cies. However, the vacancies associated with Co or H2 pressure is required to achieve HDO conversion
Ni are considerably more active than those associated of DBF, similar to the HDS conversion of DBT.
with the MoS2 alone. The vacancy concentration is In the case of BF, at least one adsorption mode may
thought to be a function of H2 and H2 S concentration. involve a bond via the C2 –C3 bond at the HYD site.
Then, the presence of H2 O and O-containing com- This would yield 2,3-dihydroBF as an intermediate
pounds in the feed can change the catalyst structure, as which will be subsequently converted to ethylphenol.
well as the geometry of the vacancy if an unsufficient Direct oxygen extrusion from BF (yielding styrene)
amount of S-donating species is present in the feed. may be favored by -bonded adsorption via an oxy-
This was confirmed in the study on the deactivation gen atom at the hydrogenolysis site. This is difficult to
of sulphided NiMo/Al2 O3 catalysts during hydropro- confirm experimentally because of the rapid removal
E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190 181
of the styrene product. Similarly, in the case of DBF, to a greater proportion of hydrogenolysis sites on the
direct oxygen extrusion yielding non-hydrogenated latter compared to that on the NiMo/Al2 O3 catalyst.
products can arise from -bonded adsorption via an Kallury et al. [92] attributed the difference between
oxygen heteroatom. After the ring opening, the oxygen the reaction paths of o-dihydroxy benzene and that
may still remain attached to the active site presumably of m- and p-dihydroxybenzenes to different modes of
as part of an OH group. Then, a non-hydrogenated adsorption on the catalyst surface. Thus, in the case
product is released after OH elimination. The fully of o-dihydroxybenzene, the O–O distance is similar
and/or partially hydrogenated products arise from as the Mo–Mo distance.
-bonded adsorption of the aromatic rings with the Interaction of bio-oil compounds with catalysts is
HYD sites. In this case, adsorption of the reactant less documented. In this regard, studies published by
has to compete with that of H2 at the same site. It is Delmon et al. [71,83,90,91,103] represent the most
believed that such adsorption is more favorable for valuable source of information. They include 4MA,
the formation of single-ring products during the HDO DES and GUA as well as H2 O, which may be formed
of DBF than that from -bonded adsorption. in large quantities. Special attention was paid to the
The involvement of two distinct sites, i.e., one hy- effects of H2 S and H2 O on the catalyst activity as well
drogenolysis and the other HYD site, is being used [117].
to interpret the HDO of phenols. The former yields In the case of 4MA, a small conversion to
the parent aromatic as a primary product, whereas the methylethyl benzene occurred even in the absence
HYD sites give cycloalkanes, presumably via an alco- of catalysts [83]. The CoMo catalyst supported on
hol and/or ketone which cannot be detected because ␥-Al2 O3 and/or carbon was much more active than
of rapid disappearance. These two routes arise from unsupported CoMoS and that supported on SiO2 . This
different reactant adsorption on the catalyst surface. is evidenced by the k4MEA in Table 15. The differ-
While studying the HDO of phenols over sulphided ence was attributed to a more efficient dispersion of
CoMo/Al2 O3 catalysts, Gevert et al. [68] proposed MoS2 on supports such as ␥-Al2 O3 and carbon. This
that the aromatic product can arise from -bonding implies that the active sites for these reactions are
adsorption through the oxygen atom, whereas the situated on metal sulphides. This is supported by the
hydrogenated product arose from -bonding ad- absence of any poisoning effect of NH3 on the HDO
sorption through the benzene ring. The former is of 4MA. Apparently, the carbonyl group is adsorbed
affected by 2,6-dialkyl substitution. Thus, Gevert via its -electrons. Two tentative routes were pro-
et al. [100] showed that a triple anion vacancy site posed. The first mode is based on the adsorption of
is required for a -bonding of 2,6-dimethylphenol, the carbonyl carbon on a nucleophilic sulphur atom.
compared to a double vacancy site required for that The reaction then proceeds via addition of a proton
of 3,5-dimethylphenol. In the case of -bonding, the to negative oxygen, and subsequently, the addition
surface dimensions of both reactants were the same; of a hydrogen atom to carbon. The second tentative
therefore, steric effects cannot explain temperature mechanism proposed by Delmon et al. [90] involves
effects on HYD (k2 ) shown in Fig. 30. Gevert et al. reaction of the carbonyl group with an activated mo-
[100] suggested that differences in delocalized res- bile hydridic species, so-called hydrogen spillover.
onance effects arising from the position of methyl The interaction of cyclohexanone with the surface of
groups influenced the HYD route. The involvement Ni3 S2 proposed by Olivas et al. [118] is shown in Fig.
of two distinct sites was supported by the strongly 35. They suggested that a similar interaction with SH
inhibiting effect of H2 S on the path, giving aromatic groups attached to Mo may occur as well.
products while hardly affecting the path yielding the In the case of the DES, except for the potassium
hydrogenated product [68]. This observation is in modified catalyst, all ␥-Al2 O3 -supported catalysts
agreement with the results published by Laurent and had the highest decarboxylation and de-esterification
Delmon [71] for CoMo/Al2 O3 catalysts. However, activities, whereas SiO2 supported catalysts had a
these authors observed that the inhibiting effect of negligible decarboxylation activity but relatively im-
H2 S on the NiMo/Al2 O3 catalyst differed from that on portant de-esterification activity. Carbon-supported
CoMo/Al2 O3 catalyst. The difference was attributed catalyst had a moderate decarboxylation activity and
182 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
that an accurate account of these effects can only be of real feeds involved a gas oil derived from heavy oil
obtained for single reactants and simple mixtures. In by thermal hydrocracking. The content of S, N and O
the case of real feeds, these effects are rather complex in the feed was 3.69, 0.39 and 0.44 wt.%, respectively.
and only some general trends may be established from The work was performed at 673 K over a sulphided
the relative removal of heteroatoms. All available in- CoMo/Al2 O3 catalyst and H2 pressure of 13.7 MPa
formation shows that HDS is much greater than HDN [121]. In this case, the O content was determined
and HDO. However, the relative rates of HDN and by neutron activation analysis and acid numbers by
HDO are not clearly established. KOH titration. The increase in the acid number with
On account of a low O content, little information is increasing Mo content of the catalyst, in Fig. 36, was
available on the HDO of fractions derived from con- attributed to the conversion of neutral BFs to acidic
ventional crudes. One of the first studies on the HDO phenols. This study was expanded to include simul-
Fig. 36. Effect of MoO3 content on O removal (1a) and acid No. (1b) during HDO of gas oil [121].
184 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
Table 22
for hydroprocessing a CDL containing 820 ppm,
Activation energies (kJ/mol) for HDN, HDS and HDO of
coal-derived feed [126] 1420 ppm and 1.24 wt.% of S, N and O, respectively,
between 548 and 673 K and 4.7 MPa of H2 . Relative
Feed Catalyst EHDN EHDS EHDO
activities of these sulphides for HDS, HDN and HDO
Illinois 6 Co–Mo 49 31 48 are shown in Fig. 39. It is evident that RuS2 exhibited
Illinois 6 Ni–W 50 36 35 the highest activity for HDN followed by HDS and
Illinois 6 Ni–Mo 48 42 38
Black Thunder Co–Mo 33 45 19
HDO, whereas for the other sulphides, the following
Black Thunder Ni–W 40 41 20 order was established; HDS>HDO>HDN. Song et al.
[128] used several NiMo catalysts of similar chem-
ical composition but different mean pore diameter
their low reactivity. Perhaps, the most detailed study and identified pore size ranges which are optimal for
on the relative removal of heteroatoms from CDLs HDO of heavy feed derived from coal.
was published by Davis et al. [126], who used two Hydroprocessing of the feeds derived from various
naphtha samples derived from Illinois 6 and Black oil shales was reviewed in detail by Landau [104], who
Thunder coals. The activation energies obtained be- established the following order for overall heteroatom
tween 493 and 673 K are shown in Table 22. In every removal: HDS>HDO>HDN. This is supported by
case, HDS was greater than HDN and HDO. However, the results published by Holmes and Thomas [129].
among five cases involving different combinations of These authors hydroprocessed the crude obtained
catalyst and naphtha, in one case HDN>HDO, two from Paraho oil shale over a sulphided NiMo/Al2 O3
cases HDN≈HDO and for two cases HDO>HDN. catalyst at about 673 K. The feed contained 0.7, 2.0
These results [126] can be used to illustrate the effects and 2.2 wt.% of S, O and N, respectively. In this case,
of catalyst type and feed origin on the relative removal almost complete removal of S was achieved, whereas
of heteroatoms. In the subsequent study, Davis et al. that of O and N was 95 and 80%, respectively. Under
[127] used the second row transition metal sulphides similar conditions, Afonso et al. [26] obtained about
Fig. 39. Simultaneous HDS (䊏), HDN (䉱) and HDO (䊉) of coal-derived naphtha over metal sulfides [127].
186 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
was more evident during the subsequent upgrading in of the latter is now understood to the point that the
the presence of an NiWO/Al2 O3 catalyst at 598 K and production of commercial fuels can be undertaken,
about 18 MPa of H2 , although significant O removal providing that the economics are favorable. Much
was achieved. less information is available on the HDO of the oil
Apparently, the chemical composition of the feed shale-derived feeds, particularly, the presence of the
is one of the factors determining the relative re- carboxylic groups-containing compounds, which are
moval of S, N and O during hydroprocessing. In resistant to HDO, requires additional attention. With
fact, it may change the generally accepted order, i.e., respect to HDO, the most complex feeds are those
HDS>HDO>HDN. For example, if in a feed, the least derived from biomass. In this regard, in recent years,
reactive O-compounds (DBFs) are predominant, to- the hydroprocessing of bio-oils has been receiving
gether with more reactive N-compounds (quinolines), most of the attention.
the overall HDN may be greater than the overall Some uncertainties in establishing the order of the
HDO [106,144]. Carbazoles appear to be the least relative HDO reactivities of the O-compounds and/or
reactive N-compounds [106]. However, their boiling groups of the compounds still exist. In most cases, the
point is significantly greater than that of DBFs, sug- results published by different authors were obtained
gesting that the occurrence of the latter together with under the different experimental conditions. This pre-
more volatile quinolines in the same feed is possible. vents direct comparison of the results. Even for a sin-
Catalysts can have a significant effect on the relative gle O-compound, reactivity is influenced by the type
heteroatom removal. The study conducted in the pres- of solvent used and the reactant concentration because
ence of an in-situ produced MoS2 , published by Ting of the self-inhibiting effects. Complications arise when
et al. [145], represents one extreme. In this case, the the results obtained for single-model compounds are
following order was established: HDS≈HDNHDO. compared with those obtained for the same compounds
Another extreme is the study in which HDO was in the various mixtures because of the mutual inhibit-
about 10 times greater than HDS in the presence of a ing and poisoning effects. The activation energies for
VN catalyst [52]. some model compounds indicate that their relative re-
activities may change with temperature. The type of
catalyst is another factor influencing the relative HDO
9. Conclusions reactivities. It appears that differences in the relative
HDO reactivities reported by different authors can be
A wealth of information on the origin of O- rationalized by thoroughly evaluating the experimen-
compounds in various feeds has been established. tal parameters used. Significant efforts would be re-
The HDO of a wide range of feeds consisting of the quired to obtain additional database for establishing a
single model O-compounds, their mixtures with S- more accurate order of the relative HDO reactivities.
and N-compounds as well as real feeds have been In view of others priorities in the catalysis research, it
studied under different experimental conditions. The is not certain whether it would be worthwhile to un-
mechanism of the single O-compounds can now be dertake such a task.
described accurately including their self-inhibiting It is generally accepted that the relative removal
effects, as well as the inhibiting and poisoning effects of heteroatoms during hydroprocessing occurs in the
of S- and N-compounds, respectively. The effect of following order: HDS>HDO>HDN. This is true for
the O-compounds on HDS and HDN has been stud- the feeds in which analogous S-, O- and N-containing
ied as well. Information is available on the effect of compounds predominate. However, a different order,
the main HDO product, such as H2 O on the catalyst e.g., HDS>HDN>HDO, was also observed and can
surface, including its effect on the overall HDO, HDS be explained by the presence of the least reactive
and HDN. O-compounds (furanic rings) together with more re-
The complexity of HDO depends on the feed ori- active N-compounds (quinolines) in the feed. Such a
gin. HDO plays a minor role during the hydroprocess- situation is not unusual when boiling points of the sin-
ing of conventional crudes, whereas its role during gle O- and N-compounds are taken into consideration.
that of the CDLs is rather major. Hydroprocessing Thus, based on the boiling points, a naphtha and/or
188 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
light gas oil could have a high content of DBFs and 4MA 4-Methylacetophenone
quinolines and a low content of the carbazoles which SRC Solvent refined coal
are the least reactive N-compounds. The type of cat- XA Xanthene
alyst used can influence the relative heteroatom re- THF Tetrahydrofuran
moval as well. For example, novel catalysts, possess-
ing about 10 times higher activity for C–O bond (in
BF) hydrogenolysis than that for the C–S bond (BT) References
hydrogenolysis have been tested.
[1] E. Furimsky, Catal. Rev.-Sci. Eng. 25 (1983) 421.
The upgrading of bio-oils remains the primary area [2] D.L. Trimm, Design of Industrial Catalysts, Elsevier,
of the current interests in HDO. It is believed that Amsterdam, 1980.
breakthroughs in the catalyst development may be re- [3] P. Grange, Catal. Rev.-Sci. Eng. 21 (1980) 135.
quired to make this source of fuels more attractive. [4] V.J. Lostaglio, J.D. Carruthers, Chem. Eng. Progr. 82 (March
1986) 46.
Other than Mo, W based catalysts may be more suit-
[5] M. Absi-Halabi, A. Stanislaus, D.L. Trimm, Appl. Catal. 72
able as it is already indicated by some preliminary in- (1991) 193.
formation. For bio-oils from pyrolysis, an additional [6] P. Grange, X. Vanhaeren, Catal. Today 36 (1997) 375.
(HYD) stage may be needed because of a high aro- [7] M.J. Girgis, B.C. Gates, Ind. Eng. Chem. Res. 30 (1991)
matics content of products after a two-stage upgrad- 2021.
[8] H. Topsøe, B.S. Clausen, F.E. Massoth, in: J.R. Anderson,
ing. It is evident that a complete HDO is the main ob- M. Boudart (Eds.), Hydrotreating Catalysis Science and
jective of the second upgrading stage conducted under Technology, Vol. 11, Springer, Berlin, 1996.
typical hydroprocessing conditions. It is believed that [9] E. Furimsky, F.E. Massoth, Catal. Today 52 (1999) 381.
the feed pretreatment (e.g., dewatering, desalting etc.) [10] E. Furimsky, F.E. Massoth, Catal. Today 17 (1993) 537.
[11] E. Furimsky, Appl. Catal. 171 (1998) 177.
prior to upgrading is not receiving adequate attention,
[12] D. Sajkowski, S.T. Oyama, in: Proc. Am. Chem. Soc. Div.
although it may slow down catalyst deactivation. The Petr. Prepr. 199th ACS Meeting, Boston, April 1990.
production of fuels from bio-oils may be looked at [13] J. Speight, Chemistry and Technology of Petroleum, 1991,
from another point of view. Thus, it may be at least a Marcel Dekker, New York.
partial answer to the environmental problems associ- [14] W.K. Robbins, Am. Chem. Soc. Div. Petr. Chem. Prepr. 43
(1998) 137.
ated with the storage and disposal of biomass. [15] J.A. Howard, Adv. Free Rad. Chem. 4 (1972) 49.
[16] L. Petrakis, R.G. Ruberto, D.C. Young, B.C. Gates, Ind. Eng.
Chem. Proc. Des. Dev. 22 (1983) 292.
10. List of symbols [17] L. Petrakis, D.C. Young, R.G. Ruberto, B.C. Gates, Ind. Eng.
Chem. Proc. Des. Dev. 22 (1983) 298.
[18] D.V. Grandy, L. Petrakis, D.C. Young, B.C. Gates, Nature
ANI Anisol 308 (1984) 175.
BF Benzofuran [19] S.S. Katti, D.W.B. Westerman, B.C. Gates, T. Youngless, L.
BT Benzothiophene Petrakis, Ind. Eng. Chem. Proc. Des. Dev. 23 (1984) 773.
CAT Catechol [20] D.W. Grandy, L. Petrakis, C.-L. Li, B.C. Gates, Ind. Eng.
Chem. Proc. Des. Dev. 25 (1986) 40.
CDL Coal-derived liquid
[21] C.-L. Li, Z. Xu, B.C. Gates, L. Petrakis, Ind. Eng. Chem.
CUS Coordinatively unsaturated sites Proc. Des. Dev. 24 (1985) 92.
DBF Dibenzofuran [22] W.H. McClennen, H.L.C. Meuzelaar, G.S. Metcalf, G.R. Hill,
DBT Dibenzothiophene Fuel 62 (1983) 1422.
DEC Decanoic acid [23] E. Furimsky, Fuel Proc. Technol. 6 (1982) 1.
[24] G. Bett, T.G. Harvey, T.W. Matheson, K.C. Pratt, Fuel 62
DES Diethylsebacate (1983) 1445.
EDEC Ethyldecanoate [25] C.E. Rovere, P.T. Crisp, J. Ellis, P. Bolton, Fuel 69 (1990)
EDS Exxon donor solvent 1099.
GUA Guaiacol [26] J.C. Afonso, M. Schmal, J.N. Cardoso, Ind. Eng. Chem. Res.
HDM Hydrodemetallization 31 (1992) 1045.
[27] M. Novotny, J.W. Strand, S.L. Smith, D. Wiesler, F.J.
HDN Hydrodenitrogenation Schwende, Fuel 60 (1981) 213.
HDO Hydrodeoxygenation [28] M.M. Boduszynski, R.J. Hurtubise, H.F. Silver, Anal. Chem.
HYD Hydrogenation 54 (1982) 3375.
E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190 189
[29] D. Beckman, D.C. Elliott, Can. J. Chem. Eng. 63 (1985) 99. [67] H. Weigold, Fuel 61 (1982) 1021.
[30] R. Maggi, B. Delmon, Biomass & Bioenergy 7 (1994) 245. [68] B.S. Gevert, J-E. Otterstedt, F.E. Massoth, Appl. Catal. 31
[31] R. Maggi, B. Delmon, in: A.V. Bridgwater (Ed.), Adv. in (1987) 119.
Thermoch. Biomass Conversion, Vol. 2, Blackie, London, [69] E. Furimsky, J.A. Mikhlin, D.Q. Jones, T. Adley, H.
1992, p. 1086. Baikowitz, Can. J. Chem. Eng. 64 (1986) 982.
[32] D.C. Elliot, in: R.P. Overend, T.A. Milne, L.K. Mudge (Eds.), [70] S.R. Kirby, C. Song, H.H. Schobert, Catal. Today 31 (1996)
Fundamentals of Thermochemical Biomass Conversion, 121.
Butterworths, Kent, 1984, p. 1002. [71] E. Laurent, B. Delmon, Ind. Eng. Chem. Res. 32 (1993) 2516.
[33] R. Maggi, B. Delmon, in: A.V. Bridgwater (Ed.), [72] R. Wandas, J. Surygala, E. Sliwka, Fuel 75 (1996) 687.
Thermochemical Biomass Conversion, Vol. 2, 1992, Elsevier, [73] C.-L. Li, Z.-R. Xu, Z.-A. Cao, B.C. Gates, AIChE J. 31
New York. (1985) 170.
[34] S.W. Benson, Thermochemical Kinetics, Wiley, New York, [74] M.W. Vogelzang, C.-L. Li, G.C.A. Schuit, B.C. Gates, L.
1968. Petrakis, J. Catal. 84 (1993) 170.
[35] D.R. Stull, E.F. Westrum, G.C. Sinke, The Chemical [75] M.J. Girgis, B.C. Gates, Ind. Eng. Chem. Res. 33 (1994)
Thermodynamics of Organic Compounds, Wiley, New York, 2301.
1969. [76] C.W. Curtis, J.L. Pellegrino, Energy & Fuels 3 (1989) 160.
[36] M.C. Edelman, M.K. Maholland, R.M. Baldwin, S.W. [77] C.M. Lee, C.N. Satterfield, Energy & Fuels 5 (1991) 163.
Cowley, J. Catal. 111 (1988) 243. [78] D.I. Kim, D.T. Allen, Ind. Eng. Chem. Res. 33 (1994) 2942.
[37] E. Furimsky, C.H. Amberg, Can. J. Chem. 10 (1976) 1507. [79] S. Chon, D.T. Allen, AIChE J. 37 (1991) 1730.
[38] E. Furimsky, Erdöl und Kohle 36 (1983) 518. [80] L. Artok, O. Erbatur, H.H. Schobert, Fuel Proc. Technol. 47
[39] A.R. Katritzky, C.W. Rees, in: Comprehensive Heterocyclic (1996) 153.
Chemistry, Vol. 4, 1984, Pergamon Press, New York. [81] F.P. Petrocelli, M.T. Klein, Fuel Sci. Technol. Int. 5 (1987)
[40] E. Furimsky, Appl. Catal. 6 (1983) 159. 25.
[41] R.J. Angelici, Polyhedron 16 (1997) 3073. [82] J. Shabtai, N.K. Nag, F.E. Massoth, J. Catal. 104 (1987) 413.
[42] K.V.R. Chary, K.S. Rama Rao, G. Muralidhar, P. Kanta Rao, [83] E. Laurent, B. Delmon, Appl. Catal. 109 (1994) 77.
Carbon 29 (1991) 478. [84] E. Laurent, A. Centeno, B. Delmon, in: B. Delmon, G.F.
[43] J. Yang, F.E. Massoth, Appl. Catal. 34 (1987) 215. Froment (Eds.), Catalyst Deactivation, Stud. Surf. Sci. Catal.,
[44] T.L. Cable, F.E. Massoth, M.G. Thomas, Fuel Proc. Technol. Vol. 88, Elsevier, Amsterdam, 1994, p. 573.
10 (1985) 105. [85] M. Ferrari, A. Centeno, C. Lahousse, R. Maggi, P. Grange, B.
[45] E. Furimsky, Ind. Eng. Chem. Prod. Res. Dev. 22 (1983) 31. Delmon, Am. Chem. Soc. Div. Petr. Chem. Prepr. 43 (1998)
[46] E. Furimsky, Ind. Eng. Chem. Prod. Res. Dev. 22 (1983) 34. 94.
[47] C. Kordulis, A. Gouromihou, A. Lycourghiotis, C. [86] J. B-son Bredenberg, M. Huuska, J. Räty, M. Korpio, J. Catal.
Papadopoulo, H.K. Mataralis, Appl. Catal. 67 (1990) 39. 77 (1982) 242.
[48] M. Bartok, G. Szollosi, J. Apjok, React. Kinet. Catal. Lett. [87] M. Huuska, J. Rintala, J. Catal. 94 (1985) 230.
64 (1998) 21. [88] J. B-son Bredenberg, M. Huuska, P. Toropainen, J. Catal. 120
[49] K. Kreuzer, R. Kramer, J. Catal. 167 (1997) 391. (1989) 401.
[50] C.-L. Lee, D.F. Ollis, J. Catal. 87 (1984) 325. [89] P. Toropainen, J.B. Bredenberg, Appl. Catal. 52 (1989) 57.
[51] C.N. Satterfield, S.H. Yang, J. Catal. 81 (1983) 335. [90] E. Laurent, B. Delmon, Appl. Catal. 109 (1994) 97.
[52] S. Ramanathan, S.T. Oyama, J. Phys. Chem. 99 (1995) 16365. [91] A. Centeno, E. Laurent, B. Delmon, J. Catal. 154 (1995) 288.
[53] C.-L. Lee, D.F. Ollis, J. Catal. 87 (1984) 332. [92] R.K.M.R. Kallury, W.M. Restivo, T.T. Tidwell, D.G.B.
[54] B. Dhandapani, T. St. Clair, S.T. Oyama, Appl. Catal. 168 Bookock, A. Crimi, J. Douglas, J. Catal. 96 (1985) 535.
(1998) 219. [93] T.P. Eskay, P.F. Britt, A.C. Buchanan, Am. Chem. Soc. Div.
[55] H. Abe, A.T. Bell, Catal. Lett. 18 (1993) 1. Fuel Chem. Prepr. 41 (1996) 1084.
[56] T. Kabe, W. Qian, A. Ishihara, J. Phys. Chem. 98 (1994) 912. [94] M. Ratcliff, F. Posey, H. Li cChum, Am. Chem. Soc. Div.
[57] S. Krishnamurthy, S. Panvelker, Y.T. Shah, AIChE J. 27 Fuel Chem. Prepr. 32 (1988) 249.
(1981) 994. [95] A. Vuori, A. Helenius, J. B-son Bredenberg, Appl. Catal. 52
[58] P.A. Hertan, F.P. Larkins, W.R. Jackson, Fuel Proc. Technol. (1989) 41.
10 (1985) 121. [96] T.-R. Viljava, A.O.I. Krause, Appl. Catal. 135 (1996) 317.
[59] V. LaVopa, C.N. Satterfield, Energy & Fuels 1 (1987) 323. [97] V. LaVopa, C.N. Satterfield, Chem. Eng. Sci. 43 (1988) 2175.
[60] M.J. Girgis, B.C. Gates, Ind. Eng. Chem. Res. 33 (1994) [98] V. Lavopa, C.N. Satterfield, J. Catal. 110 (1998) 375.
1098. [99] S.-L. Li, S.S. Katti, B.C. Gates, J. Catal. 85 (1984) 256.
[61] L.D. Rollman, J. Catal. 46 (1977) 243. [100] S.B. Gevert, M. Eriksson, P. Eriksson, F.E. Massoth, Appl.
[62] E.O. Odebunmi, D.F. Ollis, J. Catal. 80 (1983) 56. Catal. 117 (1994) 151.
[63] E.O. Odebunmi, D.F. Ollis, J. Catal. 80 (1983) 65. [101] E. Laurent, B. Delmon, in: B. Delmon, G.F. Froment (Eds.),
[64] E.O. Odebunmi, D.F. Ollis, J. Catal. 80 (1983) 76. Catalyst Deactivation, Stud. Surf. Sci. Catal., Vol. 88, Elsevier,
[65] N.P. Samchenko, N.V. Pavlenko, React. Kinet. Catal. Lett. 18 Amsterdam, 1994, p. 460.
(1981) 155. [102] S.J. Hurf, M.T. Klein, Ind. Eng. Chem. Fundam. 22 (1983)
[66] E.-J. Shin, M.A. Keane, J. Catal. 173 (1998) 450. 426.
190 E. Furimsky / Applied Catalysis A: General 199 (2000) 147–190
[103] R. Durand, P. Geneste, C. Moreau, J.L. Pirat, J. Catal. 90 [125] Y. Sato, Catal. Today 39 (1997) 89.
(1984) 147. [126] S.-L. Liaw, R.A. Keogh, G.A. Thomas, B.H. Davis, Energy
[104] M.V. Landau, Catal. Today 36 (1997) 393. & Fuels 8 (1994) 581.
[105] P. Grange, E. Laurent, R. Maggi, A. Centeno, B. Delmon, [127] A.P. Raje, S.-J. Liaw, R. Srinivansan, B.H. Davis, Appl.
Catal. Today 29 (1996) 297. Catal. 150 (1997) 297.
[106] W. Böhringer, H. Schulz, Bull. Soc. Chim. Belg. 100 (1991) [128] C. Song, K. Hanaoka, M. Nomura, Energy & Fuels 6 (1992)
831. 619.
[107] M.J. Ledoux, B. Djellouli, Appl. Catal. 67 (1990) 81. [129] S.A. Holmes, L.F. Thomas, Fuel 62 (1983) 709.
[108] G.M. Dolce, P.E. Savage, L.T. Thompson, Energy & Fuels [130] T.G. Harvey, T.W. Matheson, K.C. Pratt, M.S. Stanborough,
11 (1997) 668. Fuel 64 (1985) 925.
[109] R. Kapoor, S.T. Oyama, B. Fruhberger, J.G. Chen, J. Phys. [131] F. Goudriaan, D.G.R. Peferoen, Chem. Eng. Sci. 45 (1990)
Chem. B 101 (1997) 1543. 2729.
[110] B.C. Gates, H. Topsøe, Polyhedron 16 (1997) 3213. [132] S.B. Gevert, in: D.L. Klass (Ed.), Energy from Biomass and
[111] M. Nagai, T. Masunaga, N. Hana-oka, J. Catal. 101 (1986) Wastes XI, IGT, Chicago, 1988.
284. [133] S.B. Gevert, B.W. Anderson, S.P. Sanfqvist, S.G. Jaras, M.T.
[112] C. Moreau, C. Aubert, R. Durand, N. Zmimita, P. Geneste, Tokarz, Energy & Fuels 4 (1990) 78.
Catal. Today 4 (1988) 117. [134] E.G. Baker, D.C. Elliott, ACS Symp. Ser. 376 (1988) 228.
[113] Y. Yoshimura, T. Sato, H. Shimada, N. Matsubayashi, A. [135] B. Gevert, in: A.V. Bridgwater (Ed.), Adv. in Themochem.
Nishijima, Appl. Catal. 73 (1991) 55. Biomass Conversion, Vol. 2, Blackie, London, 1992, p. 1424.
[114] Y. Yamamoto, F. Kumata, F.E. Massoth, Fuel Proc. Technol. [136] R.W. Thring, J. Breau, Fuel 75 (1996) 795.
19 (1998) 253. [137] M.C. Samolanda, W. Baldauf, I.A. Vasalos, Fuel 77 (1998)
[115] S. Kasztelan, A. Wambeke, L. Jalowiecki, J. Grimblot, J.P. 1667.
Bonelle, J. Catal. 124 (1990) 12. [138] B.S. Gevert, J.-E. Otterstedt, Biomass 14 (1987) 173.
[116] Y. Okamoto, A. Maezawa, T. Imanaka, J. Catal. 120 (1989) [139] D.C. Elliott, E.G. Baker, in: Energy from Biomass and
29. Wastes X, Washington, 1986, Elsevier, New York, p. 765.
[117] E. Laurent, C. Pierret, O. Keymeulen, B. Delmon, in: A.V. [140] E.G. Baker, D.C. Elliott, in: A.V. Bridgwater, J.L. Kuster
Bridgwater (Ed.), Adv. in Thermochem. Biomass Conversion, (Eds.), Research in Thermochemical Biomass Conversion,
Vol. 2, Blackie, London, 1992, p. 1403. Elsevier, London, 1988, p. 883.
[118] A. Olivas, V. Petranovskii, J. Cruz-Reyes, S. Fuentes, Am. [141] E. Churin, R. Maggi, P. Grange, B. Delmon, in: A.V.
Chem. Soc. Div. Petr. Chem. Prepr. 43 (1998) 51. Bridgwater, J.L. Kuester (Eds.), Research in Thermochemical
[119] H. Knözinger, H. Krietenbrink, H.-D. Müller, W. Schultz, Biomass Conversion, Elsevier, London and New York, 1988,
in: Proc. 6th Int. Congr. Catal., London, 1976, Vol. 1, Chem. p. 896.
Soc., London, 1977, p. 183. [142] J.D. Rocha, C.A. Luengo, C.E. Snape, Renew. Energy 9
[120] H. Topsøe, H. Topsøe, F.E. Massoth, J. Catal. 119 (1989) (1996) 950.
252. [143] J. Gagnon, S. Kaliaguine, Ind. Eng. Chem. Res. 27 (1988)
[121] E. Furimsky, Fuel 57 (1978) 494. 1783.
[122] E. Furimsky, AIChE J. 25 (1979) 306. [144] C.K. Groot, V.H.J. de Beer, R. Prince, M. Stolarski, W.S.
[123] D.K. Dalling, G. Haider, R.J. Pugmire, J. Shabtai, W.E. Niedzwiedz, Ind. Eng. Chem. Prod. Res. Dev. 25 (1986)
Hull, Fuel 63 (1984) 525. 522.
[124] Y. Yoshimura, K. Hayamizu, T. Sato, H. Shimada, A. [145] P.-S. Ting, C.W. Curtis, D.C. Cronauer, Energy & Fuels 6
Nishijima, Fuel Proc. Technol. 16 (1987) 55. (1992) 511.