Push-pull mechanism of hydrodenitrogenation over silica-supported MoP, WP, and MoS2 hydroprocessing catalysts

Article abstract of DOI:10.1006/jcat.2002.3658

The mechanism of liquid-phase catalytic hydrodenitrogenation at 3.1 MPa on silica-supported molybdenum phosphide, MoP/SiO₂, and tungsten phosphide, WP/SiO₂, was studied using a series of pentylamines of different structures. The reactivity pattern suggested that removal of nitrogen occurred primarily by an E2 elimination mechanism involving acidic and base sites on the catalyst surfaces in a push-pull process. Infrared spectroscopy and temperature-programmed reaction studies of ethylamine indicated that alkyl ammonium species formed on Bronsted acid sites were intermediates in the reaction. Similar results were obtained with a reference MoS₂/SiO₂ sample tested at the same conditions. This suggested that sulfur was probably present on the active surface and assisted in the removal of sulfur.

Full text of DOI:10.1006/jcat.2002.3658

Journal of Catalysis 210, 116–126 (2002)
doi:10.1006/jcat.2002.3658
Push–Pull Mechanism of Hydrodenitrogenation over Silica-Supported
MoP, WP, and MoS2 Hydroprocessing Catalysts
∗
∗
∗, ,1
P. Clark, X. Wang, P. Deck,† and S. T. Oyama †
∗
Environmental Catalysis and Materials Laboratory, Department of Chemical Engineering, Virginia Polytechnic
Institute and State University, Blacksburg, Virginia 24061-0211; and †Department of Chemistry,
Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061-0211
Received January 9, 2002; revised April 26, 2002; accepted May 2, 2002
conditions at which it is run, such as hydrogen pressure,
residence time, and the presence of H2S, but also the cata-
lyst may be complex and may possess multiple functional-
ities, such as hydrogenation and hydrogenolysis sites (10).
It is likely that the mechanism and rate of reaction will
depend ultimately on the nature of its interaction with
the reacting molecule. The most well known mechanism
of denitrogenation is the Hoffmann degradation of amines
(11), but there are a number of other possibilities (12–14)
such as nucleophilic substitution (SN1, SN2), elimination
(E1, E2), and metallacycle or metal alkyl formation path-
ways (Scheme 1). The mononuclear pathways (SN1 and E1)
start with the generation of a carbocation by quaterniza-
tion of an amine followed by cleavage of ammonia. Nucle-
ophilic substitution occurs by attack of a nucleophile (HS−)
on the carbenium ion (the original α-carbon), whereas
elimination occurs by abstraction of a proton by the
nucleophile (from the original β-carbon). The binuclear
pathways (SN2 and E2) begin with the formation of the
quaternized ammonium compound as in the mononuclear
case, but without scission of ammonia. Nucleophilic substi-
tution occurs by attack of a nucleophile on the α-carbon
whereas elimination (E2) occurs by attack of a base on a
proton attached to a β-carbon to form an olefin. The path-
ways involving reactions with metals (C–N oxidative addi-
tion and metal alkyl formation) require activation of the
α-carbon to the nitrogen (15–17). These pathways likely
occur with metallic catalysts such as Rh, Pd, or Ru but
are not preferred with the phosphide or sulfide catalysts
considered here. To ascertain the operative mechanism
(SN1, E1, SN2, or E2) on the catalysts, use was made of
the method of Breysse (12–14) which employed a series
of pentylamines of different structure. The original work
was carried out in the vapor phase, and we have adapted
the method to employ more realistic conditions of liquid
phase and high pressure (3.1 MPa). Comparison is made
to a simple sulfide deposited on a noninteracting support,
MoS2/SiO2, to avoid complications due to promoters and
acidic supports.
The mechanism of liquid-phase catalytic hydrodenitrogenation
at 3.1 MPa on silica-supported molybdenum phosphide, MoP/SiO2,
and tungsten phosphide, WP/SiO2, was studied using a series of
pentylamines of different structures. The reactivity pattern sug-
gested that removal of nitrogen occurred primarily by an E2 elim-
ination mechanism involving acidic and base sites on the cata-
lyst surfaces in a push–pull process. Infrared spectroscopy and
temperature-programmed reaction studies of ethylamine indicated
that alkyl ammonium species formed on Brønsted acid sites were
intermediates in the reaction. Similar results were obtained with
a reference MoS2/SiO2 sample tested at the same conditions. This
suggested that sulfur was probably present on the active surface
c
and assisted in the removal of sulfur.
ꢀ 2002 Elsevier Science (USA)
Key Words: hydrodenitrogenation; mechanism; tungsten phos-
phide; molybdenum sulfide; pentylamine.
INTRODUCTION
Transition-metal phosphides have recently been re-
ported to be effective hydroprocessing catalysts, extending
the range of materials known to be active for this class of
reactions. In particular MoP (1–4), WP (5, 6), Co2P (7),
and Ni2P (8) have been shown to give high conversions
in hydrodesulfurization (HDS), and especially hydrodeni-
trogenation (HDN), of model petroleum feedstocks. Lit-
tle is known about the nature of these catalytic materi-
als, and studies to elucidate their behavior are of practical
and fundamental interest. We have undertaken the present
investigation to understand the mechanism by which
the phosphide catalysts carry out the HDN reaction of
alkylamines.
The subject of HDN on sulfides, carbides, and nitrides
has recently been comprehensively reviewed by Prins (9).
Elucidation of the mechanism of HDN is complicated for
several reasons. Not only is the reaction affected by the
1 To whom correspondence should be addressed. E-mail: oyama@
vt.edu.
116
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

PUSH–PULL MECHANISM OF HYDRODENITROGENATION
117
The sulfide samples used for hydroprocessing and O2
chemisorption were sulfided directly from the oxide within
the catalytic reactor using the same procedure.
All gases were supplied by Air Products. Hydrogen, he-
lium, and carbon monoxide were all 99.999% pure and were
passed through a water trap (Alltech) before contacting
the sample. Specialty mixtures, including 0.5% O2/He, 30%
N2/He, and 10% H2S/H2, were used as received.
The total dynamic CO uptake was measured by passing
pulses of CO (5.6 µmol) in a stream of 40 µmol s−1 helium
over samples at room temperature (RT). Measurements
were performed ex situ on air-exposed samples reduced to
723 K for 2 h. Oxygen uptakes were measured in the same
mannerexceptthatthechemisorptiontemperaturewasthat
of dry ice/acetone (195 K) and the O2 uptake was multiplied
by a factor of 2 to account for dissociative adsorption on the
surface. Surface area measurements by N2 adsorption were
performed with a Micromeritics 2000 sorption unit.
X-ray diffraction (XRD) spectra were collected with a
Scintag XDS-2000 X-ray diffractometer using Ni-filtered
Cu Kα (λ = 0.1541 nm) radiation. Scans were collected at a
rate of 0.035◦ 2θ s−1 and in increments of 0.03◦ 2θ.
SCHEME 1. Mechanisms and intermediates in C–N bond cleavage.
EXPERIMENTAL
The mechanism of HDN was studied by hydrode-
Supported phosphides were prepared by incipient wet- nitrogenation of solutions containing 2000 wppm N
ness impregnation of silica with aqueous metal phosphate (wppm = parts per million by weight) as n-pentylamine
precursors, followed by controlled reduction in flowing hy- (C5H13N, Acros, 99%), tert-pentylamine (Acros, 98%), or
drogen. The preparation, characterization, and catalytic ac- neo-pentylamine (TCI) dissolved in tetradecane (Fisher,
tivityofthesilica-supportedphosphideswerereportedinan 99%). These reactions were carried out at 3.1 MPa as a func-
earlier paper (18). Briefly, stoichiometric quantities of am- tion of temperature for each catalyst. The pentylamine feed
monium paramolybdate, (NH4)6Mo7O24 · 4H2O (Aldrich, solutions also contained 3000 wppm S as dimethyldisulfide
99%), orammoniummetatungstate, (NH4)6W12O39 · xH2O (C2H6S2, Aldrich, 99%), as well as 2000 wppm n-octane
(Aldrich, 90%), were combined with ammonium phos- (C8H18, Aldrich, 99%) as an internal standard.
phate, (NH4)2HPO4 (Aldrich, 99%), in distilled water, and
The basis used for the calculation of catalytic reaction
the solution was used to impregnate the silica support. results was the percent of reactant converted to a given
Quantities of reagents were used to give weight loadings product, where all of the detected reactant and product
of 13% MoP/SiO2 and 20% WP/SiO2, corresponding to species are counted as 100%. HDN was defined as the con-
molar loadings of 1.16 mmol metal/g SiO2. The moist paste version of the amine to hydrocarbon products and so did
obtained after impregnation was calcined in air at 773 K not include condensation (transalkylation) products. The
for 6 h. The resulting precursor phosphates were reduced carbon balances on the amine reactants and their products
to phosphides in quartz U-tube reactors (1 cm o.d.) us- were typically 90% 10%. The balance was low because
ing linear temperature ramps in flowing hydrogen. Hydro- only the liquid-phase products were analyzed. Gas–liquid
gen flow was set at 650 µmol s−1 g−1 starting material (e.g., equilibria were considered in the liquid samples after col-
300 cm3(NTP) min−1 H2 flow for a 0.300 g sample) for all lection to help account for the volatility of the C5 hydrocar-
syntheses. The MoP/SiO2 required a final reduction temper- bons and amines at ambient conditions (19). Small amounts
ature of 850 K and the WP/SiO2 required a final tempera- of light products were found in dry-ice temperature con-
ture of 1000 K. After reduction, the phosphides were cooled densate from the hydrogen (vapor) effluent, but it was not
to room temperature under 65 µmol s−1 helium and typi- possible to accurately quantify this material because com-
cally were passivated progressively by flowing 78 µmol s−1 plications arose from carryover of the liquid-phase solvent
of 0.1% O2/He overnight, followed by 13 µ mol s−1 of 0.5% tetradecane in the gas stream.
O2/He for 2 h, and then allowing ambient air to diffuse into
the reactor tube for 24 h before collecting the samples.
The composition of hydroprocessing liquids was iden-
tified with a Hewlett–Packard 5890A gas chromatograph
Samples of 5.6 wt% MoS2/SiO2 corresponding to (equipped with a CP Sil 5B column) on samples collected
0.368 mmol metal/g SiO2 were prepared by sulfiding at 2–3-h intervals. Steady state was typically achieved after
MoO3/SiO2 for 2 h in a stream of 10% H2S/H2 at 678 K. about 60 h of reaction time at a given condition. Reaction

118
CLARK ET AL.
products were identified through matching of retention
times with commercially available standards, as well as
by GC/MS analysis. The GC/MS combined a Fisons Carlo
Erba 8060 series gas chromatograph, using a HP-5MS 5%
phenylmethylsiloxane stationary phase, with a VG Quat-
tro triple quadrupole mass spectrometer operated in the
electron impact–positive ion mode.
RESULTS
Preparation and Characterization of Catalysts
The catalysts were prepared and characterized as de-
scribed previously (18). Briefly, the phosphides, MoP/SiO2
and WP/SiO2, were obtained by the temperature-
programmed reduction of supported phosphate precursors.
The catalysts were shown to be single-phase materials by
31P nuclear magnetic resonance spectroscopy. The sulfide,
MoS2/SiO2, was obtained by sulfidation of a supported ox-
ide precursor. A summary of surface area and chemisorp-
tion properties is provided in Table 1.
A study of the reactivity of a series of pentylamines was
carried out to explore the effect of amine structure on the
HDN reaction. The results for each individual amine are
presented in the following.
Tostartacatalyticreaction, phosphidecatalystswerepre-
treated in 100 µmol s−1 hydrogen at 723 K and 1 atm pres-
sure for 2 h, while the sulfide catalyst was presulfided with
a 10% H2S/H2 mixture at 678 K and 1 atm pressure for 2 h.
After the pretreatment, the conditions were set to 3.1 MPa
(450 psig), a hydrogen flow rate of 100 µmol s−1 (150 cm3
NTP min−1), a liquid feed rate of 0.0833 cm3 s−1 (5 cm3
min−1), and the desired temperature. For amine reactions,
the amounts of catalysts used were 3.3 g of MoS2/SiO2,
1.4 g of MoP/SiO2, and 3.1 g of WP/SiO2. These amounts
corresponded to roughly similar quantities (56, 70, and
58 µmol, respectively) of chemisorption sites. Once started,
the individual catalysts were used for extended sets of
conditions (i.e., different feeds and temperatures). The
duration of the experiments was lengthy, requiring several
weeks, but standard conditions were repeated periodically
to verify that no deactivation occurred. Results from re-
peated conditions were similar and were averaged to obtain
the reported values.
Diffuse reflectance Fourier transform infrared (FTIR)
spectra of ethylamine in the gas phase and adsorbed on
the WP/SiO2 catalyst and the SiO2 support were mea-
sured with a Bio-Rad FTS 60A spectrometer equipped
with a Spectra-Tech HTEC-0030-103 high-temperature en-
vironmental chamber. The finely powdered catalysts were
pressed flat in the sample holder of the environmental
chamber and pretreated as for reaction. They were then
exposed to 3000 wppm S in the form of H2S in H2 at 553 K
followed by six ethylamine pulses (20 µmol) in H2. Spec-
tral acquisition consisted of 1024 scans at different temper-
atures after purging with H2. The catalyst sample data were
normalized by the background spectrum acquired using a
blank KBr sample.
n-Pentylamine
The product distributions for n-pentylamine HDN on
silica-supported MoP, WP, and MoS2 are presented in Fig. 1.
The species favored at the lowest temperature in each case
was the condensation product dipentylamine. As tempera-
ture increased, pentane appeared along with small amounts
of pentenes, pentanethiol, and tripentylamine. Total n-
pentylamine conversions at all temperatures except 513 K
increased in the order MoP/SiO2 < MoS2/SiO2 < WP/SiO2;
however, the production of hydrocarbons increased in the
order MoP/SiO2 < WP/SiO2 < MoS2/SiO2. The MoP/SiO2
catalyst (Fig. 1a) was the least active by both measures.
Comparatively, the WP/SiO2 catalyst (Fig. 1b) had a higher
propensity for condensation than MoS2/SiO2, while the
MoS2/SiO2 catalyst (Fig. 1c) had higher hydrocarbon
production.
TABLE 1
Surface Area and Chemisorption Characteristics of Fresh
and Spent Silica-Supported Catalyst Samples
CO uptakeb
Temperature-programmed desorption (TPD) of ethyl-
amine from the WP/SiO2 catalyst was carried out using H2
as the carrier gas. As with the FTIR measurements, the
sample was pretreated as for reaction and then exposed
to an H2S/H2 stream with a concentration of 3000 wppm
S. The sample was purged in H2 and the adsorption of
ethylamine was carried out in pulse mode as before. The sat-
urated sample was again purged in H2 and the temperature-
Sample
Surface area,a fresh
Fresh
Spentc
SiO2 blank
MoS2/SiO2
MoP/SiO2
WP/SiO2
91
93d
50
62
0
0
—
26
18e
50
19
10.8
a Surface area in m2 g−1 catalyst; passivated catalysts; degassed in vac-
programmed experiment was carried out using a linear uum at 273 K for 24 h.
b Uptake at room temperature in µmol g−1; samples pretreated at 723 K
in H2.
c Uptake at room temperature following hydrodenitrogenation of
temperature increase of 0.135 K s−1 to a final temperature
of 1273 K. The desorbing species from the catalyst were
monitored with a mass spectrometer during the heating.
The areas for the desorbing species were compared with
the areas from calibrated pulses of reference gases.
pentylamines.
d As oxide: MoO3/SiO2.
e O2 uptake at 78 K; sample presulfided at 678 K in 10% H2S/H2.

PUSH–PULL MECHANISM OF HYDRODENITROGENATION
119
tert-Pentylamine
The product distributions for tert-pentylamine HDN on
silica-supported MoP, WP, and MoS2 are presented in Fig. 2.
Hydrocarbons were the only products of tert-pentylamine
HDN, with olefins being preferred at lower temperatures,
and the alkane 2-methylbutane becoming more important
at higher temperatures. The order of activity of the catalysts
increased as MoP/SiO2 < WP/SiO2 < MoS2/SiO2, with the
MoP catalyst (Fig. 2a) being significantly less active than
either the WP (Fig. 2b) or MoS2 (Fig. 2c) catalysts. The rel-
ative amount of 2-methyl-2-butene was always greater than
2-methyl-1-butene on each catalyst. Di(tert-pentyl)amine
was not observed, in agreement with previous results (12),
and its absence is attributed to steric effects.
neo-Pentylamine
The product distributions for neo-pentylamine HDN
on silica-supported MoP, WP, and MoS2 are presented
in Fig. 3. The initial product formed at low tempera-
tures/conversions was di(neo-pentyl)amine. This molecule
was the only product detected on the MoP catalyst
(Fig. 3a). On WP (Fig. 3b) and MoS2 (Fig. 3c), 2,2-
dimethylpropane was observed at higher temperatures and
was the only hydrocarbon product detected. The order of
reactivity of the catalysts in terms of both total conver-
sion and the formation of hydrocarbons increased in the
order MoP/SiO2 < WP/SiO2 < MoS2/SiO2. Small amounts
of 2,2-dimethylpropanenitrile were detected in products
FIG. 1. Product distribution in the hydrogenolysis of n-pentylamine
on (a) MoP/SiO2, (b) WP/SiO2, and (c) MoS2/SiO2 catalysts: (×) conver-
sion, (ꢀ) dipentylamine, (ꢁ) n-pentane, (ꢂ) tripentylamine, (ꢃ) pentenes,
and (ꢄ) pentanethiol.
Theamountsanddistributionsofolefinsat533Karesum-
marized in Table 2, along with equilibrium concentrations
estimated by thermodynamic calculations for this temper-
ature. The predominant olefin formed on all catalysts was
1-pentene. Cis- and trans-2-pentenes were observed for the
WP/SiO2 and MoS2/SiO2 catalysts at higher temperatures
(513 K). In no cases were isopentene isomers observed.
Overall, the hydrocarbon distribution was consistent
with the occurrence of HDN by an elimination reaction fol-
lowed by dehydrogenation. However, significant amounts
of pentanethiol were also found at 493 K on the MoS2/SiO2
catalyst, demonstrating the presence of N–S exchange as
well.
TABLE 2
Olefin Production and Distribution for n-Pentylamine
HDN at 513 K
% Selectivityb
Total
Sample
olefins, %a 1-Pentene trans-2-Pentene cis-2-Pentene
c
MoP/SiO2
WP/SiO2
3
6
12
—
100
42
39
7
0
30
39
69
0
28
22
24
MoS2/SiO2
Equilibrium
a Percent of n-pentylamine converted to olefins.
b Percent of total olefin.
FIG. 2. Product distribution in the hydrogenolysis of tert-pentylamine
on (a) MoP/SiO2, (b) WP/SiO2, and (c) MoS2/SiO2 catalysts: (×) conver-
sion, (ꢀ) 2-methyl-2-butene, (ꢄ) 2-methyl-1-butene, and (ꢁ) n-pentane.
c MoP results are from 533 K. Total hydrocarbons for MoP at 513 K = 0.

120
CLARK ET AL.
FIG. 5. Fourier transform infrared spectra of ethylamine adsorbed on
WP/SiO2.
produced by WP and MoS2 containing catalysts. Thus, neo-
pentylamine is the only molecule for which a nitrile product
was observed in this study.
An overall summary of the reactivity of the MoP/SiO2,
WP/SiO2, and MoS2/SiO2 catalysts in HDN is provided in
Fig. 4. The data in that figure are extracted from Figs. 1–3
to show only conversion to denitrogenated hydrocarbon
products.
FIG. 3. Product distribution in the hydrogenolysis of neo-
pentylamine on (a) MoP/SiO2, (b) WP/SiO2, and (c) MoS2/SiO2 catalysts:
(×) conversion, (ꢀ) di(neo-pentyl)amine, (ꢁ) dimethylpropane, and
(ꢂ) 2,2-dimethylpropylnitrile.
The FTIR spectra at different temperatures of adsorbed
ethylamine on the WP/SiO2 catalyst exposed to H2S are
shown in Fig. 5. Major features are seen at 2260, 2400,
2530, 2885, 2943, 2977, 3305, and 3370 cm−1. These features
diminish in intensity as the temperature is raised from 298
to 333 K.
FTIR spectra taken at similar conditions on the SiO2 sup-
port are shown in Fig. 6. Here features are observed at 2260,
FIG. 4. Total HDN conversion of pentylamines over (a) MoP/SiO2,
FIG. 6. Fourier transform infrared spectra of ethylamine adsorbed on
(b) WP/SiO2, and (c) MoS2/SiO2 catalysts.
SiO2.

PUSH–PULL MECHANISM OF HYDRODENITROGENATION
121
The surface area of the MoS2/SiO2 catalyst (93 m2 g−1) pre-
cursor was close to that of the support (91 m2 g−1) because
its loading was low. The surface areas of the MoP/SiO2
(50 m2 g−1) and WP/SiO2 (62 m2 g−1) samples were sig-
nificantly decreased from that of the support, a common
finding for phosphorus-promoted supported catalysts (20,
21). This is presumably an effect of pore blocking by the
supported catalyst compounds as surface area decreases
with loading (3). Chemisorption quantities of O2 and CO
for the fresh catalysts were moderate and were found to de-
crease somewhat after catalytic reaction for the phosphides.
Chemisorptionmeasurementstakenaftercatalyticreaction
were carried out on recovered samples washed in hexane
and rereduced at 723 K for 2 h. The fresh MoP/SiO2 had a
largerCOuptake(50 µmolg−1)thantheWP/SiO2 (19µmol
g
−1). This probably was the result of the lower synthesis
FIG. 7. Temperature-programmed desorption traces of ethylamine
from WP/SiO2: (a) desorption of ethylamine, (b) desorption of ethylene,
and (c) desorption of ammonia.
temperature required for MoP/SiO2 (850 K) than WP/SiO2
(1000 K), which reflects the relative reducibilities of the
Mo and W precursors. The dispersion of MoP/SiO2 (4.9%)
was consequently larger than that of WP/SiO2 (2.0%). As
noted earlier, the reactivity work presented here is based
on approximately equal chemisorption sites loaded in the
catalytic reactor. The sites are titrated by the adsorption of
CO at RT on the phosphides and O2 at low temperature for
the sulfides. These are reasonable techniques. On the phos-
phides our previous characterization work (18) showed that
the CO uptake is close to the expected metal site density ob-
tained from the crystallite sizes. Moreover, the conversion
in HDS and HDN was invariant with loading when equal
sites were loaded in the reactor (3). On the sulfides care-
ful studies have clearly shown that on unpromoted MoS2
the low-temperature chemisorption of O2 tracked linearly
with activity (22–26). This is another reason for choosing a
simple system like MoS2/SiO2 for comparison.
It is generally recognized that HDN is more difficult than
HDS in hydroprocessing because of the higher strength (by
12–38 kJ mol−1) of C–N bonds versus C–S bonds (27). The
high activity of the phosphide catalysts for HDN was un-
usual (18) and prompted investigation of the mechanism
of HDN. A method described by Breysse’s group (12–
14) utilizing a series of pentylamines of different struc-
ture was adapted to high-pressure, liquid-phase conditions
in our study. The amines consisted of n-pentylamine, tert-
pentylamine, and neo-pentylamine and their structure and
attributes are described in Table 3. For each amine the
table lists the number of hydrogen atoms on the α- and
β-carbons, the degree of steric hindrance, and the relative
stability of carbocations and carbanions formed on removal
of the amine group.
2882, 2948, 2978, and 3340 cm−1, and again these diminish
in intensity as the temperature is increased. There was no
difference in the spectra with H2S pretreatment, indicating
that no acid sites are formed by the action of H2S on the
SiO2.
The TPD traces of ethylamine from WP/SiO2 exposed
to H2S are shown in Fig. 7. The figure displays the mass
spectrometer signals from three masses representing ethy-
lamine (mass = 30), ethylene (mass = 28), and ammonia
(mass = 15) as a function of temperature. The assignment
of these masses was checked by examination of secondary
masses. For example, the fragmentation pattern of ethy-
lamine has the following distribution: mass 30 (100), mass
28 (32), mass 44 (18), mass 45 (17), mass 27 (15), and mass
15 (11), where the numbers in parentheses indicate rela-
tive intensity. Because of the high intensity for mass 30, this
was the primary signal monitored for ethylamine. In order
to distinguish it from ethane, which also has mass 30, the
signal was compared to that of mass 45. The topmost trace
(Fig. 7, curve a), corresponding to the desorption of the par-
ent compound ethylamine, shows a relatively sharp peak at
380 K and a broad feature with a maximum at 485 K. The
other traces (Fig. 7, curves b and c) show corresponding
signals which are partially due to the fragmentation of the
parent compound. However, the traces also show distinct
peaks at 613 K for ethylene and 650 K for ammonia, indicat-
ing the separate production of these species. Integration of
the area under these peaks and quantitation by comparison
to the signals obtained from injected standards give desorp-
tion amounts of 52 µmol g−1 for ethylene and 56 µmol g−1
for ammonia.
The expected relative reactivities of the pentylamines
are summarized in Table 4, as described by Cattenot et al.
(12). neo-Pentylamine contains no β-hydrogen atoms and
is sterically hindered, so is expected to be the least reac-
DISCUSSION
The ex situ chemisorption characteristics and BET sur- tive molecule in the series. tert-Pentylamine has the great-
faceareasofthecatalystsamplesaresummarizedinTable1. est number of β-hydrogen atoms and forms the most stable

122
CLARK ET AL.
TABLE 3
Relative Properties of Pentylamines for HDN Reaction
Number of
α-hydrogens
Number of
β-hydrogens
Steric
hindrance
Carbocation
stability
Carbanion
stability
Reactant
Structure
n-Pentylamine
2
0
2
8
Low
Low
High
Low
tert-Pentylamine
High
High
neo-Pentylamine
2
0
Med
Med
Med
carbocation counterpart, so it is the most reactive molecule
Infrared spectroscopy studies with ethylamine give in-
in the series. n-Pentylamine has an intermediate number sight about how the reaction proceeds. Ethylamine is the
of β-hydrogen atoms, is the least sterically hindered, and smallest member of the alkylamine family, having both α-
is expected to have intermediate activity. Carbocations of and β-hydrogens, and is a good probe molecule for the
neo-pentylamine and n-pentylamine are expected to have mechanism. In this study measurements were conducted
similar stability.
on the WP/SiO2 catalyst because of its high activity. The
The complete conversion and selectivity results were pre- catalyst was pretreated in an H2 stream with 3000 wppm
sented in Figs. 1–3. The reactions of n-pentylamine and neo- S (the same sulfur level used during reaction) to approx-
pentylamine were predominantly to condensation prod- imate the actual catalytic surface conditions. The FTIR
ucts, whilethoseoftert-pentylamineweretodenitrogenated measurements indicated that ethylamine is adsorbed on
molecules.
the surface of the WP/SiO2 as an ethylammonium species.
The conversions to hydrocarbons (HDN) of each catalyst Previous studies on Mo2C and MoS2/SiO2 gave similar re-
for each of the three amines are plotted in Fig. 4. The trend sults (28). This was deduced from the characteristic bands
for the three pentylamine HDN reactions over MoP/SiO2, of this species at 2260, 2400, 2530, 2885, 2943, 2977, 3305,
WP/SiO2, and MoS2/SiO2 catalysts is for increased reactiv- and 3370 cm−1 (29–31). The bands in the midrange region
ity in the order neo < n < tert. Examination of the structural (2885–2977 cm−1) are due to C–H bond symmetric and
properties of the amines (Table 3) shows no correlation of asymmetric vibrations. The weak bands at low wavenumber
this order of reactivity with steric hindrance, suggesting that belong to the parent ethylamine compound (2260 cm−1
)
an SN2 pathway is not predominant. Similarly, there is no and an ethylammonium species (2400–2530 cm−1), while
correlation with the stability of carbocation or carbanion the doublet at high wavenumber (3305–3370 cm−1) is also
species, suggesting that prior scission of the amine group due to the ethylammonium species. The ethylamine parent
as in an SN1 or E1 mechanism does not occur. However, is likely associated with the SiO2 support. The IR bands on
there is a direct relationship between the order of reactiv- theSiO2 arealsofoundinthelow(2260cm−1)andmidrange
ity and the number of β-hydrogens, and this suggests that regions (2882–2978 cm−1).
C–N bond scission occurs primarily by an E2 mechanism on
TPD studies of the adsorbed ethylamine species were un-
the catalysts (Table 4). Of course, the preferred E2 path- dertaken to understand the manner in which the reaction
way is not available for neo-pentylamine and this molecule proceeds. Again, the studies employed the active WP/SiO2
is likely to react by a metal-type reaction (Scheme 1). This catalyst pretreated with 3000 ppm S. The TPD signals from
is discussed in more depth later.
the mass spectrometer are shown in Fig. 7. Ethylamine
TABLE 4
Relative Reactivities of Pentylamines According to Possible Reaction Mechanisms
Number of
Reactant
β-Hydrogens
SN1
SN2
High
Very low
Very low
E1
E2
n-Pentylamine
tert-Pentylamine
neo-Pentylamine
2
8
0
Very low
High
Very low
Very low
Very high
None
Medium high
High
None

PUSH–PULL MECHANISM OF HYDRODENITROGENATION
123
Table 2). At higher conversions the 1-pentene intermedi-
ate was hydrogenated to form n-pentane or isomerized by
double bond migration to form cis- and trans-2-pentenes.
Importantly, no isopentenes were observed, indicating that
carbeniumionintermediatessuchasthoseformedbySN1or
E1 pathways were not formed. The lack of skeletal isomer-
ization of the hydrocarbon suggests that the double bond
migration step also occurred by a push–pull-type mecha-
nism (Scheme 3). Again, in this scheme the surface struc-
tures are pictorial representations.
Thermodynamic calculations show that 2-pentene is fa-
vored over 1-pentene and trans-2-pentene is favored over
cis-2-pentene. In the case of n-pentylamine the cis–trans
distribution in the catalyst effluent was skewed toward 1 : 1,
suggesting a kinetic control over the double bond migration
step. From studies of relative stability of allylic species (33)
the preference toward cis-2-pentene may indicate that the
intermediate has partial anion character.
In the case of tert-pentylamine, the products of reaction
were exclusively olefins and paraffins. The formation of
2-methyl-2-butene was preferred over the formation of 2-
methyl-1-butene over each catalyst. This result shows pref-
erence for the thermodynamically favored triply substi-
tuted olefin product compared to the kinetically favored
product having three times more available β-hydrogen
atoms (6 : 2) for E2 elimination. If E1 elimination does not
occur, this suggests that the methylbutenes can isomerize
readily. However, it is likely in this case, because of the sta-
bility of the tertiary carbenium formed, that the SN1 mech-
anism contributes to the reaction.
The reactions of neo-pentylamine fall into a different
category from those of n-pentylamine and tert-pentylamine
because the molecule does not have any β-hydrogens, mak-
ing standard E1 and E2 elimination reactions impossible.
Similarly, there is no γ -elimination as indicated by the ab-
sence of methylbutenes (34). As far as nucleophilic sub-
stitutions are concerned, the SN1 route is unlikely because
the intermediary primary carbenium ion is unstable, while
the SN2 pathway is blocked by steric hindrance. There is a
SCHEME 2. Push–pull mechanism in the HDN of an alkylamine.
desorbs molecularly in a relatively sharp peak at 380 K
and a broader peak at 485 K (Fig. 7, curve a). These fea-
tures probably correspond to physisorbed and chemisorbed
ethylamine, respectively. The other TPD traces show that
part of the adsorbed ethylamine decomposes to release am-
monia and ethylene in equal proportions, but in a noncon-
certed manner. Ethylene (Fig. 7, curve b) appears first at
613 K, while ammonia (Fig. 7, curve c) desorbs later at
650 K. A possible sequence of steps for the transforma-
tion is depicted in Scheme 2. In this scheme we have taken
the surface of the phosphide catalyst to be populated with
sulfur species which provide Brønsted acidity in the form of
sulfhydril groups, as well as nucleophilic centers. The exact
structure of these groups is uncertain, and the forms shown
inScheme2shouldbeconsideredillustrative, asotherpossi-
bilities exist. However, the main features of the mechanism
are otherwise consistent with the observations. The amine is
first protonated on a Brønsted site on the surface to adsorb
as an alkylammonium species. This species then undergoes
a β-hydride elimination by attack by a surface basic center.
This is exactly an E2 elimination (Scheme 1) with the role
of the surface now explicitly indicated, and results in the
release of ethylene, as found in the TPD experiments. The
eliminated ammonia interacts with the coordinatively un-
saturated site on the surface and is released in a subsequent
step.
Another material that has received attention as an al-
ternate HDN catalyst is molybdenum carbide (32). Inter-
estingly, in similar experiments using ethylamine TPD it
was found that for Mo2C (also presulfided) the adsorbed
intermediate released ammonia first and then ethylene,
the reverse order of the WP catalyst (28). This suggests
that the nature of carbide and phosphide catalyst surfaces
are quite distinct. The reactivity patterns of n-pentylamine
and tert-pentylamine are consistent with the mechanism for
WP/SiO2. In particular, for n-pentylamine it is likely that
1-pentene was the primary reaction product as it was the
major olefin product formed at low temperatures (Fig. 1,
SCHEME 3. Push–pull mechanism of olefin isomerization.

124
CLARK ET AL.
possibility of hydride attack to displace the amine group, as evant, though, that alumina has been reported as a catalyst
the existence of hydride species associated with metal cen- for denitrogenation reactions of primary amines leading to
ters has been suggested before (35) as charged species in olefins(45). Theauthorsofthestudyonthealuminacatalyst
spillover on neutral silica supports (36) and organometallic found that methyl substitution at the α-carbon increased
HDN reactions involving lithium aluminum hydride (37). the amine reaction rate, while methyl substitution at the
But again, this route appears unlikely because the hydride β-carbon decreased the rate. This is in complete agreement
is likely to be held strongly. Finally, a homolytic reaction with our findings. Thus, acid–base chemistry alone can ex-
pathway can be considered involving H radicals (38, 39). plain results for paraffinic amines. However, good activity
This probably does not occur because of the lack of skele- in hydroprocessing reactions is correlated not only with hy-
tal isomerization and β-scission products.
drogenolysis reactions but also with hydrogenation ability.
Themorelikelyscenariointhecaseof neo-pentylamineis The origin of the need for hydrogenation activity is related
activation of an alpha carbon by one of the metal-centered totherequirementofsaturationofboththeα-andβ-carbon
reactions noted in Scheme 1. The aziridium-type interme- atoms in order to allow the C–N hydrogenolysis reaction to
diate is particularly likely. This would be followed by a se- take place. It was on this basis that Nelson and Levy (11)
ries of hydrogenation and reductive elimination steps to proposed that a Hoffmann-type E2 mechanism is responsi-
produce the neo-pentane or the 2,2-dimethylpropylnitrile ble for the C–N bond cleavage reaction. Thus the activity of
species. Nitriles have been reported before in the reactions catalysts toward unsaturated molecules in real feedstocks,
of primary amines, where nitrile formation is favored by such as quinolines, is limited by the ability of the catalyst
lower pressures (40) and the absence of hydrogen (41, 42). for hydrogenation of the C6 ring. Our results confirm in
In this case, the reactive channel involving a metal-centered a new way that C–N bond cleavage is dominated by E2
reaction is likely opened up at the higher temperatures of elimination. Furthermore, our interpretation of results in
reaction needed to convert neo-pentylamine.
conjuction with acid–base chemistry is in substantial agree-
Comparison of the product distributions for the three ment with the suggestion of Topsøe et al. that the active
amines shows that a significant amount of condensation centers of hydrodenitrogenation catalysts involve Brønsted
products and thiols are found with n-pentylamine and neo- acid sites (46).
pentylamine. Thus, substitution is an important pathway
In this study it was found that MoP was the least reac-
for these molecules, especially at low temperature. In con- tive in the pentylamine reactions, while WP and MoS2 had
trast, no condensation or thiol products occur for the tert- similar activity. In contrast, in earlier work on quinoline
pentylamine HDN, indicating that elimination and hydro- HDN (18) it was shown that the conversions of MoP (31%)
genation are the only relevant reactions. In fact, the E1 and WP (46%) were higher than that of MoS2 (11%). It
mechanism is possible on tert-pentylamine because of the has been found in the literature that MoS2 is active for
stability of the carbocation. This point is highlighted by 22% hydrogenolysis reactions, while it is not as active for hydro-
conversion of tert-pentylamine in a blank run at 513 K. This genation reactions. Our finding that MoS2/SiO2 was active
was the only significant background reaction noted in this for C–N bond hydrogenolysis reactions, but was relatively
investigation.
inactive for quinoline HDN, seems to support this estab-
The results of this study are different from the results re- lished conclusion. The higher activity of Mo and W phos-
ported for bulk sulfides by Cattenot et al. (13) because ours phides than the MoS2/SiO2 in quinoline HDN indicates that
were obtained at high pressure. Hence, we have a greater they have a distinct surface composition that gives rise to
propensity for hydrogenation in this study, resulting in pen- a unique reactivity. As seen in Schemes 2 and 3, our pic-
tanes as well as pentenes. Interestingly, earlier results for ture of the catalyst involves an amphoteric surface in which
NiMo/Al2O3 operating at 2.0 MPa yielded no condensa- HDN elimination reactions operate by a push–pull mech-
tion products (14). In particular, we operate at the same anism. The amine is envisioned to adsorb from a sulfhydril
temperature (523 K), and with sulfur, but also find 2,2- Brønsted acid site by proton transfer. A nearby bridging
dimethylpropane and condensation products in the efflu- sulfide group provides the basic site necessary for the elim-
ent. This could be due to the presence of liquid solvent or ination reaction. The bridging sulfide can be viewed as the
to the nature of the catalysts.
conjugate base of the sulfhydril acid site. The sulfhydril site,
Our results for MoS2/SiO2 resembled those of bulk MoS2 as pictured, is intimately associated with a coordinatively
used by Cattenot et al. (12). In particular, both had good unsaturated site. Thus, measurements of unsaturated sites
activity for neo-pentylamine conversion and had similar by CO and O2 uptake correlate indirectly to the active acid–
product distributions for n-pentylamine conversion, includ- base sites but are effective nonetheless.
ing di(n-pentyl)amines. The presence of acid sites alone
Our conclusion that the HDN of alkylamines proceeds
does not lead to active hydroprocessing catalysts, as demon- preferentially by an E2 mechanism is supported by re-
strated by the inactivity of alumina and silica–alumina sup- sults on heterocyclic nitrogen compounds with sulfides (47).
ports, acting alone, in HDN reactions (20, 43, 44). It is rel- It is found, for example, that 2-picoline is more reactive

PUSH–PULL MECHANISM OF HYDRODENITROGENATION
125
than pyridine and 2-methylpiperidine is more reactive than and MoS2/SiO2 catalysts. Studies with ethylamine to probe
piperidine, suggesting higher HDN activity for molecules the mechanism by infrared spetroscopy and TPD suggested
with greater numbers of β-hydrogen atoms.
that the reaction involved formation of adsorbed alkylam-
It has been suggested in the literature that the essential monium species and reorganization of bonds by a push–pull
competing reaction for E2 elimination of amines involves mechanism involving acid and base sites on the surface.
nucleophilic substitution of sulfur upon amine groups (12–
14, 48). The resulting thiol is then reported to react quickly
by direct hydrogenolysis with hydrogen, forming H2S and a
saturated hydrocarbon. However, this does not properly ex-
plain the case of aliphatic thiol decomposition on MoS2, in
which 83% of the product of desulfurization of n-butylthiol
was found to be olefinic and butadiene was a major product
in the decomposition of tetrahydrothiophene (49). Thus, it
is likely that thiols also decompose primarily by E2 elimi-
nation. It has also been reported that HDS of cyclopropyl-
methylthiol occurs by a free radical mechanism (50), but
again, this does not explain the selectivity toward olefins
noted here.
In the case of benzylamine or isoquinoline HDN reac-
tions reported in the literature, HDN proceeds without
prior hydrogenation of the C6 ring, but the process is slower
(occurs at higher temperature) than in the case of aliphatic
amines. This suggests that an alternate mechanism, such as
ACKNOWLEDGMENT
The authors are indebted to the Department of Energy (DOE) Office of
Basic Energy Science, Grant DE-FG02-96ER14669, for financial support.
REFERENCES
1. Li, W., Dhandapani, B., and Oyama, S. T., Chem. Lett. 207 (1998).
2. Stinner, C., Prins, R., and Weber, Th., J. Catal. 191, 438 (2000).
3. Oyama, S. T., Clark, P., Teixeira da Silva, V. L. S., Lede, E. J., and
Requejo, F. G., J. Phys. Chem. 105, 4961 (2001).
4. Phillips, D. C., Sawhill, S. J., Self, R., and Bussell, M. E., J. Catal. 207,
266 (2002).
5. Clark, P., Li, W., and Oyama, S. T., J. Catal. 200, 140 (2001).
6. Stinner, C., Prins, R., and Weber, Th., J. Catal. 202, 187 (2001).
7. Robinson, W. R. A. M., van Gestel, J. N. M., Kora´nyi, T. I., Eijsbouts,
S., van der Kraan, A. M., van Veen, J. A. R., and de Beer, V. H. J.,
J. Catal. 161, 539 (1996).
8. Wang, X., Clark, P., and Oyama, S. T., J. Catal., in press.
9. Prins, R., Adv. Catal. 46, 399 (2001).
one of the metal-mediated routes (metal alkyl, aziridinium 10. Prins, R., Jian, M., and Flechsenhar, M., Polyhedron 16(18), 3235
(1997).
ring), is available for HDN in cases where the β-carbon is
not fully saturated (10, 43).
Our studies are carried out at high pressure, in the liq-
11. Nelson, N., and Levy, R. B., J. Catal. 58, 485 (1979).
12. Cattenot, M., Portefaix, J. L., Afonso, J., Breysse, M., Lacroix, M., and
Perot, G., J. Catal. 173, 366 (1998).
uid phase, with a sulfur cofeed to achieve realistic condi-
tions. The presence of sulfur likely alters the composition
of the surface, possibly resulting in the formation of a sur-
face phosphosulfide. We have indicated this in our schemes.
We believe that the nature of this surface sulfide is dif-
ferent from that of a pure sulfide because the underlying
bulk structure and morphology are different. This has been
demonstrated on supported carbides and nitrides, which
also form surface sulfides, but with considerably higher ac-
tivities than supported sulfides (51, 52). The same thing
probably happens with Mo and W phosphides, which un-
like the sulfides do not have a layered structure (18) and
so take on a globular morphology. This has been demon-
strated recently with elegant electron microscopy work on
MoP/SiO2 (4). The spherical morphology leads to the for-
mation of more exposed surface atoms than can be obtained
13. Portefaix, J. L., Cattenot, M., Guerriche, M., and Breysse, M., Catal.
Lett. 9, 127 (1991).
14. Portefaix, J. L., Cattenot, M., Guerriche, M., Thivolle-Cazat, J., and
Breysse, M., Catal. Today 10, 473 (1991).
15. Laine, R. M., Catal. Rev.—Sci. Eng. 25, 459 (1983).
16. Chisholm, M. H., Polyhedron 16, 3071 (1997).
17. Weller, K. J., Fox, P. A., Gray, S. D., and Wigley, D. E., Polyhedron 16,
3139 (1997).
18. Clark, P., Wang, X., and Oyama, S. T., J. Catal. 207, 256 (2002).
19. Hadden, S. T., and Grayson, H. G., Hydrocarbon Process 40, 207
(1961).
20. Kraus, H., and Prins, R., J. Catal. 170, 20 (1997).
21. Lopez Cordero, R., Esquivel, N., Lazaro, J., Fierro, J. L. G., and Lopez
Agudo, A., Appl. Catal. 48, 341 (1989).
22. Tauster, S. J., Pecoraro, T. A., and Chianelli, R. R., J. Catal. 63, 515
(1980).
23. Tauster, S. J., and Riley, K. L., J. Catal. 67, 250 (1981).
24. Bachelier, J., Duchet, J. C., and Cornet, D., J. Phys. Chem. 84, 1925
(1980).
with the raftlike structures in sulfides, which expose inac- 25. Bodrero, T. A., Bartholomew, C. H., and Pratt, K. C., J. Catal. 78, 253
(1982).
tive basal planes. This can potentially lead to an increase
in site densities for the phosphides and represents a unique
opportunity for these new materials.
26. Zmierczak, W., MuraliDhar, G., and Massoth, F. E., J. Catal. 77, 432
(1982).
27. Katzer, J. R., and Sivasubramanian, R., Catal. Rev.—Sci. Eng. 20, 155
(1979).
28. Schwartz, V., Teixeira da Silva, V., and Oyama, S. T., J. Mol. Catal. A:
CONCLUSIONS
Chem. 163, 251 (2000).
29. Parillo, D. J., Adamo, A. T., Kokotailo, G. T., and Gorte, R. J., Appl.
Catal., A 67, 107 (1990).
30. Zeroka, D., Jensen, J. D., and Samuels, A. C., J. Mol. Struct. 465, 119
The reactivity of a series of pentylamines with different
structure in HDN at high pressure confirmed the predomi-
nance of the E2 elimination mechanism in the conversion of
(1999).
primary amines to hydrocarbons on MoP/SiO2, WP/SiO2, 31. Hagemann, H., and Bill, H., J. Chem. Phys. 80, (1984).

126
CLARK ET AL.
32. Ramanathan, S., and Oyama, S. T., J. Phys. Chem. 99, 16365 (1995).
33. Schleyer, P. V. R., Dill, J. D., Pople, J. A., and Hehre, W. J., Tetrahedron
33, 2497 (1977).
34. Siddhan, S., and Narayanan, N., J. Catal. 59, 405 (1979).
35. Grimblot, J., Catal. Today 41, 111 (1998).
36. Stumbo, A. M., Grange, P., and Delmon, B., Stud. Surf. Sci. Catal. 106,
225 (1997).
44. Reinhoudt, H. R., Troost, R., van Schalkwijk, S., van Langeveld, A. D.,
Sie, S. T., Schulz, H., Chadwick, D., Cambra, J., de Beer, V. H. J., van
Veen, J. A. R., Fierro, J. L. G., and Moulijn, J. A., Stud. Surf. Sci. Catal.
106, 237 (1997).
45. Hogan, P., and Pasek, J., Collect. Czech. Chem. Commun. 38, 1513
(1973).
46. Topsøe, N. Y., Topsøe, H., and Massoth, F. E., J. Catal. 119, 252
(1989).
47. Cox, K. E., and Berg, L., Chem. Eng. Prog. 56, 54 (1962).
48. Vivier, L., Dominguez, V., Perot, G., and Kasztelan, S., J. Mol. Catal.
67, 267 (1991).
49. Kolboe, S., Can. J. Chem. 47, 352 (1969).
50. Dungey, K. E., and Curtis, M. D., J. Am. Chem. Soc. 119, 842
(1997).
51. Aegerter, P. A., Quigley, W. W. C., Simpson, G. J., Ziegler, D. D., Logan,
J. W., McCrea, K. R., Glazier, S., and Bussell, M. E., J. Catal. 164, 109
(1996).
37. Weller, K. J., Fox, P. A., Gray, S. D., and Wigley, D. E., Polyhedron
16(18), 3139 (1997).
38. Startsev, A. N., J. Mol. Catal. A 152, 1 (2000).
39. Paul, J., Akpati, H., Nordlander, P., Oh, W. S., Goodman, D. W., and
Demirel, B., Stud. Surf. Sci. Catal. 106, 303 (1997).
40. Sonnemans, J., and Mars, P., J. Catal. 34, 215 (1974).
41. Lee, J. H., Hamrin, C. E., Jr., and Davis, B. H., Appl. Catal., A 111, 11
(1994).
42. Brey, W. S., Jr., and Cobbledick, D. S., Ind. Eng. Chem. 51(9), 1031
(1959).
43. Menon, R., Joo, H. S., Guin, J. A., Reucroft, P. J., and Kim, J. Y.,
Energy Fuels 10, 579 (1996).
52. McCrea, K. R., Logan, J. W., Tarbuck, T. L., Heiser, J. L., and Bussell,
M. E., J. Catal. 171, 255 (1997).