Synthesis, characterization, and thiophene desulfurization activity of unsupported γ-Mo2N macrocrystalline catalysts

Article abstract of DOI:10.1006/jcat.1998.2307

Macrocrystals of γ-Mo₂N were synthesized by temperature-programmed reaction of macrocrystalline MoO₃ and NH₃ or N₂/H₂ mixtures. X-ray diffraction analyses indicate macrocrystalline Mo₂N is an aggregate of particles in crystallographic alignment with diameters ranging from 4.6 to 18 nm, depending on synthesis conditions. Scanning tunneling microscopy (STM) supports these observed diameters. Based on diffraction data and surface area measurements, it is concluded that the particles in each sample are found with a range of diameters, each flattened in the [200] direction with either an amorphous surface phase or polycrystalline interior. The highest crystal BET surface area achieved is 44 m²/g. TGA was used to monitor the temperature-programmed reaction of MoO₃ and N₂/H₂ mixtures. The lowest reduction temperatures were observed in syntheses employing high H₂ concentrations and slow temperature ramping rates. The rate of the Mo₂N macrocrystal synthesis reaction was observed to be slower than the equivalent powder reaction, which could possibly be attributed to the effects of solid-state diffusion. The specific thiophene hydrodesulfurization (HDS) activity (units of mol/s m²) of the Mo₂N macrocrystalline catalyst at 673 K was found to be higher than the powder form by a factor of 2.5. Weight-specific HDS activities and reaction product fractions over Mo₂N macrocrystal and powder catalysts were roughly the same. An analysis of reaction products over a range of conversions indicates thiophene desulfurizes to form predominantly 1-butene with smaller amounts of other C₄ hydrocarbons also present.

Full text of DOI:10.1006/jcat.1998.2307

Journal of Catalysis 182, 136–147 (1999)
Article ID jcat.1998.2307, available online at http://www.idealibrary.com on
Synthesis, Characterization, and Thiophene Desulfurization Activity
of Unsupported � -Mo N Macrocrystalline Catalysts
2
Eric J. Markel,�,1 Susan E. Burdick,�,2
�
,3
Marion E. Leaphart II, and Kenneth L. Roberts†
�
Department of Chemical Engineering, University of South Carolina, Columbia, South Carolina 29208; and †Department of Chemical Engineering,
North Carolina A&T State University, 330 McNair Hall, Greensboro, North Carolina 27411-0001
Received June 19, 1998; revised October 2, 1998; accepted October 12, 1998
vironmental applications. � -Mo2N prepared by this method
Macrocrystals of � -Mo2N were synthesized by temperature- has been studied as a catalyst for NH3 decomposition (6,
programmed reaction of macrocrystalline MoO3 and NH3 or 7), CO hydrogenation (8), ethane hydrolysis (8, 9), hydro-
N2/H2 mixtures. X-ray diffraction analyses indicate macrocrys-
talline Mo2N is an aggregate of particles in crystallographic align-
ment with diameters ranging from 4.6 to 18 nm, depending on syn-
thesis conditions. Scanning tunneling microscopy (STM) supports
these observed diameters. Based on diffraction data and surface
This work details the previously unreported analysis of
area measurements, it is concluded that the particles in each sam-
ple are found with a range of diameters, each flattened in the [200]
denitrogenation (10, 11), and hydrodesulfurization (12, 13)
reactions. � -Mo2N powder is a highly porous, dark metallic-
gray material composed of small particles (or crystallites)
of micrometer-scaled proportions (5).
Mo2N macrocrystals for the hydrodesulfurization (HDS)
reaction of thiophene. � -Mo2N macrocrystals are relatively
small, dark metallic-gray platelets of moderately porous
direction with either an amorphous surface phase or polycrystalline
2
interior. The highest crystal BET surface area achieved is 44 m /g.
TGA was used to monitor the temperature-programmed reaction material which have visible dimensions varying from
of MoO3 and N2/H2 mixtures. The lowest reduction temperatures micrometer- to centimeter-scaled proportions. Macrocrys-
were observed in syntheses employing high H2 concentrations and talline Mo N catalysts could be useful for several reasons
2
slow temperature ramping rates. The rate of the Mo2N macrocrys-
tal synthesis reaction was observed to be slower than the equivalent
powder reaction, which could possibly be attributed to the effects of
including providing lower pressure drops in a packed bed
reactor, being more amenable to study by advanced micro-
scopic techniques such as scanning tunneling microscopy
solid-state diffusion. The specific thiophene hydrodesulfurization
(
5), and could possibly be more selective for industrial hy-
2
(
HDS) activity (units of mol/s m ) of the Mo2N macrocrystalline
drodesulfurization reactions (14). Crystalline Mo2N also
may serve as the starting point for a range of crystal
chemistries: Mo2N powder can also be readily converted to
carbides, borides, and metal using high temperature treat-
catalyst at 673 K was found to be higher than the powder form by
a factor of 2.5. Weight-specific HDS activities and reaction prod-
uct fractions over Mo2N macrocrystal and powder catalysts were
roughly the same. An analysis of reaction products over a range
of conversions indicates thiophene desulfurizes to form predomi- ments in CH4, BH3, and H2, respectively (10).
nantly 1-butene with smaller amounts of other C4 hydrocarbons
also present.
Previous work reported from this group has examined
the effects of reaction parameters such as temperature, re-
actant gas composition, and water content on the synthesis
of high surface area molybdenum nitride powders (7) and
the reactivity of the unsupported Mo2N catalysts for the
NH3 decomposition and thiophene HDS reactions (4, 12).
Hydrogen lean mixtures of N2 and H2 were observed to
�c 1999 Academic Press
INTRODUCTION
-Mo2N powders have been synthesized previously by
�
several groups using temperature-programmed reaction of
MoO3 and NH3 (1–3) or N2/H2 mixtures (4, 5). The high spe-
cific surface area of the Mo2N powders could make them
particularly promising catalysts for many industrial and en-
2
produce mixtures of low surface area (10 m /g) material
composed of MoO2, Mo2N, and Mo for the temperature-
programmed reaction (7). This result combined with sup-
porting temperature ramping experimental data indicates
that at temperatures between 573 and 873 K MoO2 is a
1
Current address: Exxon Polymer Research Center, 5200 Bayway possible intermediate of the temperature-programmed re-
Drive, Baytown, TX 77520.
action. MoO2 and molybdenum oxynitride have previously
been reported as possible intermediates in the formation of
2
Current address: Department of Chemical Engineering, Texas A&M
University, College Station, TX 77843-3122.
3
Mo N powders for the temperature-programmed reaction
To whom correspondence should be addressed. Fax: (336)334-7904.
E-mail: kroberts@ncat.edu.
2
systems containing MoO3 and NH3 (1).
136
0021-9517/99 $30.00
Copyright �c 1999 by Academic Press
All rights of reproduction in any form reserved.

HYDRODESULFURIZATION OF THIOPHENE OVER � -Mo2N CATALYSTS
137
The hydrodesulfurization of thiophene over Mo2N pow- in a manner such that each intermediate reaction step is
ders has been previously reported by this group to pro- also topotactic. In this way, structural symmetry between
duce butane as one of the initial products of the thiophene the molybdenum lattice points exists throughout the entire
desulfurization reaction at 673 K (12). Significant amounts reaction. Bertrand and Dufour (15) studied the topotactic
of butadiene and butene products and small amounts of isothermal reduction of MoO3 to MoO2 in flowing H2 at
2
tetrahydrothiophene were also reported to form during the 100 Torr (1 Torr = 133.3 N/m ). Low temperatures (723–
desulfurization of thiophene. Sulfided Mo2N catalysts were 743 K) were used to avoid complete reduction of the ox-
observed to have much higher area specific activities for ides to Mo metal and water-free H2 was used because slight
the thiophene desulfurization reaction compared with sup- traces of water have been shown to enhance MoO3 subli-
ported sulfided cobalt/molybdenum catalysts. Mo2N cata- mation by the formation of volatile hydrates (16):
lysts were not observed to be significantly affected by a
MoO3 + H2 = MoO2 + H2O.
[1]
combination of sulfiding for 4 h in a 10% H2S in H2 mixture
and subsequent kinetic analysis for 12–18 h at 673 K as ob-
served from subsequent X-ray diffractometry and Raman
spectroscopy analyses. At temperatures above 773 K, some
MoS2 was observed as a result of the sulfiding process.
The surface areas of Mo2N catalyst powders were varied
Using the selected area electron diffraction it was shown
that the following structural relationships exist during
the topotactic reduction: [100]MoO2 k [010]MoO3, [122]
¯
¯
MoO2 k [100]MoO3, and [122]MoO2 k [001]MoO3. The au-
thors reported an activation energy of 19 � 2 kcal/mol for
this reduction.
2
from 6.0 to 108 m /g using a range of temperature rates and
a variety of preparation conditions. The area specific HDS
activity of the catalysts was observed to increase with signif-
icant decreases in surface area. Promotion of specific HDS
activity for the low surface area catalysts was attributed to
the detected presence of MoN in the catalyst samples.
Hillis, Kemball, and Roberts (17) studied the isothermal
reduction of low surface area MoO2 in flowing, atmospheric
H2 at temperatures from 543 to 788 K to form moderate
surface area molybdenum metal:
MoO2 + 2H2 = Mo + 2H2O.
[2]
REVIEW
The reaction product was identified by X-ray diffraction
as a mixture of Mo metal and the low surface area MoO2
starting material. At 723 K, the reduction kinetics were such
that a period of 12 h was required for a 1% conversion to
Mo. At equilibrium, it is expected that all of the molybde-
num dioxide be converted to the metal (18). The specific
surface area of the MoO2/Mo product mixtures increased
Volpe and Boudart (1) were the first to report the pro-
duction of high surface area � -Mo2N in temperature pro-
grammed reactions of MoO3 in flowing NH3. Subse-
quently, it was shown that ammonia decomposes during
temperature-programmed synthesis and that high surface
areas could also be achieved using N2/H2 mixtures as feed
2
2
linearly from 2 m /g to a final value of 50 m /g. Possible
topotactic relationships of the reactants and products were
not addressed.
(
4, 5). The reaction requires a slow temperature ramping
rate (0.6 K/min from 298 to 980 K) and a high gas space
�
1
velocity (typically 150,000 h ) through a packed bed of
2
powdered MoO3 to produce surface areas up to 170 m /g.
Solid structures. MoO3 forms a crystalline solid with an
These reaction conditions have been found to lead to MoO3 ReO3-type layered structure and an orthorhombic Bravais
reduction temperaturesaslowas620K. Reactionswith high lattice (0.392 nm, 1.394 nm, 0.366 nm) (19). The layered
space velocity, low heating rate, and low synthesis temper- planes are composed of corner sharing MoO6 octahedra
atures produce the desired high surface area (3, 4).
which are parallel to the (010) plane of the primitive cell.
The MoO3 reduction reactions which produce high sur- The Mo–O bond distance varies from 0.188 to 0.245 nm
face areas are termed topotactic because the � -Mo2N prod- in each octahedron (20). Only weak physical interac-
uct has a well defined three-dimensional crystallographic tions bind adjoining MoO3 layers. MoO2 has a mono-
orientation relative to the parent MoO3 crystallite. Specif- clinic structure, with ao = 0.5584 nm, bo = 0.4842 nm, co =
�
0
ically, selected area electron diffraction data indicate that 0.5608 nm, and b = 120 59 (20). The Mo–O bond length
the (010) planes of MoO3 are parallel to the (100) planes varies from 0.19 to 0.21 nm in MoO2; the Mo–Mo bond
of product � -Mo2N. X-ray diffraction and electron diffrac- length is 0.248 nm. In � -Mo2N, the Bravais lattice is face
tion data show that the powdered � -Mo2N product con- centered cubic, with ao = 0.4165 nm (21). Each molybde-
sists of micrometer-sized porous platelet aggregates of 3- num atom is octahedrally surrounded by six nonmetal-
1
2
1
2
1
2
to 4-nm-diameter crystallites in crystallographic alignment lic sublattice sites. One Mo atom is present at , ,
(
1). Thus, the platelet morphology of MoO3 is preserved in and nitrogen is statistically distributed over the octahe-
topotactic Mo2N. dral interstitial sites of the metallic sublattice as posi-
Because the reduction of MoO3 to � -Mo2N is topotactic, tions , 0, ; , , 0; 0, , ; 1, , ; , 1, ; and , , 1 as
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
the removal of oxygen and infusion of nitrogen must occur required to maintain a stoichiometric ratio. Using these

1
38
MARKEL ET AL.
parameters, the separation of Mo–Mo atoms in � -Mo2N and BET surface area analysis. The fresh, unused catalysts
is 0.2945 nm, and the Mo-N separation is 0.208 nm (20).
were passivated and exposed to air before analysis. Fresh
macrocrystalline catalysts were also examined using SEM
and STM analyses. The “used” catalysts were treated in a
EXPERIMENTAL
2
.7% thiophene in hydrogen mixture and heated from room
temperature to 673 K in the flowing mixture at 0.125 K/s.
The flowing thiophene/hydrogen mixture was maintained
at 0.5 ml/s over the bed of catalyst particles with variable
superficial space velocity for a period of 24 h. Following the
kinetic analysis, the catalysts were cooled to room temper-
ature and exposed to air.
Catalyst synthesis. Crystalline Mo2N was synthesized
from macrocrystals of MoO3. These starting macrocrys-
tals were produced from MoO3 powder (Johnson Matthey,
9
9.999% ) using high temperature sublimation. Approxi-
mately 100 g of MoO3 powder was loaded into a silica quartz
semicylindrical boat, placed into a sealed 2-inch-diameter
silica combustion tube and heated in a large Lindberg fur-
nace under flowing nitrogen gas. As the MoO3 powder
X-ray diffraction analysis. A Rigaku Model D-max-B
single-axis diffractometer equipped with a Cu source oper-
ating at 20 mA and 30 kV was used for both powder and
macrocrystal analyses. The detector was equipped with a
monochromator for the rejection of all but the CuK� line.
All data were compared to JCPDS values for identification
�
sublimated at temperatures above 500 C, product MoO3
macrocrystals were deposited at the opposite end of the
combustion tube and maintained at room temperature. Af-
ter the complete sublimation of the powder MoO3, the fur-
nace was cooled and the MoO3 macrocrystalline product
was removed.
(
20). Diffraction peak width data were analyzed to deter-
mine particle size using the Scherrer equation (23).
A special holder was designed to perform analyses of
macrocrystals using the single-axis goniometer. The holder
allowed the (100), (010), (001), and (111) planes of Mo2N
macrocrystal samples to be aligned in the direction parallel
to the sample plane in the goniometer (Fig. 1), thus allow-
ing one-dimensional analyses of Mo2N macrocrystals in a
powder diffractometer. Mo2N macrocrystal samples were
mounted into the holder using double-sided tape with mi-
nor alignment corrections performed with the samples in
place in the goniometer.
Approximately 25.0 g of MoO3 macrocrystals was loaded
into a quartz boat and placed in a combustion tube and
connected to an NH3 (Matheson, 99.9999% ) source. The
transparent yellow crystals obtained exhibit the flat platelet
habit on MoO3 and were typically more than 1 cm long, 3–
5
mm wide, and less than 200 �m thick. Similar results were
obtained using 75% H2/25% N2 mixtures (Matheson, N2,
9
9
9.9995% , and H2, 99.9995% ). The sample was heated to
80 K at 0.6 K/min in NH3 flowing at 0.001 m /min. At the
3
end of the temperature program, the sample was removed
from the furnace and cooled without exposure to oxygen.
The cooled sample is air-sensitive and must be passivated
before removal to the atmosphere. Passivation of the Mo2N
macrocrystals was accomplished by stopping the ammonia
flow and allowing air to contact the sample through 0.5 m
of 0.635-cm-i.d. tube for 48 h. Previous work in this group
has found that an amorphous MoO2 surface layer forms on
the surface of passivated � -Mo2N powders from diffuse re-
flectance FTIR and indicated by XRD measurements (22).
Previous work has shown that substantial losses in surface
area accompany the passivation procedure (1).
Molybdenum nitride powder was synthesized using me-
thods previously reported (4). Approximately 0.1 g of
MoO3 powder was loaded into a quartz tube fitted with
a porous ceramic frit and reacted in flowing NH3 or N2/H2
at atmospheric pressures, using gas flow procedures and
the temperature program methods described above. After
the reaction was completed, the reactant gas was allowed
to continue flowing and the Mo2N powder was cooled to
room temperature. After cooling, the reactant gas flow was
halted and the solid product was passivated using air that
contacted the product sample through 0.5 m of 0.635-cm-i.d.
tubing for 24 h.
FIG. 1. Orientation of Mo2N macrocrystal in goniometer stage for
one-dimensional X-ray diffraction studies. The characteristic flat platelet
morphology of MoO3 is retained in the Mo2N macrocrystal with the (200)
Characterization. Catalysts were characterized both
before and after thiophene reaction using X-ray diffraction plane of Mo2N parallel to the basal (010) plane of MoO3.

HYDRODESULFURIZATION OF THIOPHENE OVER � -Mo2N CATALYSTS
139
Surface area analysis. BET surface area analysis was with a flame ionization detector (FID) and a Graphpac
performed using a Micromeritics 2700 dynamic adsorption GC column (Alltech, Inc.) for the analysis of HDS reac-
analyzer. The N2/He mixtures used in adsorption measure- tion products. After allowing a 20-min stabilization period
ments were prepared in the laboratory by mixing high pu- at 673 K, the analytical system automatically sampled and
rity gases (Matheson, N2, 99.9995% , and He, 99.9995% ) in analyzed the reaction products every 20 min over a 12-h
a Sierra electronic dual-channel mass flow controller. An period. Afterward, the reactant mixture flow rate was in-
integrating thermal conductivity detector was used to de- creased from 0.5 ml/s to 1.0, 2.0, 3.33, and 6.66 ml/s to de-
termine test gas composition during adsorption/desorption. crease residence time and approach differential conversion
Adsorption was carried out at 77 K.
operation. Several analyses were obtained at each elevated
flow rate for each catalyst. Comparison studies were car-
ried out using the procedures mentioned above using Mo2N
powder catalysts.
Scanning electron microscopy. A Hitachi Model S2500
Delta microscope was used for SEM analyses. Crystalline
Mo2N samples were mounted on SEM stubs for analysis by
evaporating a small portion of solution containing a dilute
aqueous suspension of the dispersed crystals. Solution con-
centrations over several orders of magnitude were tested
to optimize particle density on the stub.
RESULTS
Macrocrystal synthesis. The crystalline MoO3 starting
material has a flat needle morphology typically 1 cm long,
Scanning tunneling microscopy. Analyses were per-
formed using a Digital Nanoscope II system equipped with
a 1- by 1- by 0.3-�m scan head and platinum tip. A sin-
gle crystal of topotactic Mo2N was mounted onto the scan
head using carbon paste. The top layers of the macrocrystals
could be easily peeled away using adhesive tape to reveal a
fresh, clean surface for analysis.
3
–5 mm wide, and less than 200 �m thick. During reaction,
the transparent macrocrystals darken, starting at the edges
until the entire solid is black. Following complete reduction,
nitridation, and passivation, the product has a dark metallic
luster and is quite flexible. Gross macrocrystal morphology
appears to be unchanged.
The reduction/nitridation reaction of MoO3 macrocrys-
Thermogravimetric analysis. Mo2N syntheses were also tals with 75% H2 in N2 was investigated using thermogravi-
performed using a thermogravimetric analyzer (Perkin- metric analysis (TGA). During temperature programmed
Elmer TGA7). Typically, 25 mg of MoO3 macrocrystals or reaction at 0.6 K/min (Fig. 2), reaction commences at 650 K
powder were loaded into a platinum sample pan for analy- with a total mass loss of 28% , corresponding to that ex-
sis. A gas–liquid feed system allowed for the programmed pected for the complete conversion of MoO3 to Mo2N.
addition of high purity (>99.99% ) nitrogen, hydrogen, wa- Reduction of MoO3 powder occurs at much lower tem-
ter, helium, and a mixture of 1.0% oxygen in nitrogen for peratures. The TGA profiles do not plateau at any dis-
passivation purposes. It should be noted that careful system crete intermediate compositions. X-ray diffraction analy-
purging both before and after reaction must be exercised ses of the intermediate compositions show the presence of
in these experiments involving hydrogen in the presence small amounts of low surface area MoO2 as well as a solid
of oxygen and platinum. The analyzer was clamped shut to phase previously reported (1, 4) but not fully character-
prevent opening in the event of a power failure. It should ized with lattice planes at 205 pm. This result from TGA
be noted that hydrogen is produced by catalytic decompo-
sition of NH3 during Mo2N synthesis in NH3 (4) and that
appropriate purges are necessary in syntheses employing
NH3.
Thiophene desulfurization kinetic measurements. Thio-
phene desulfurization rates were measured in a continuous
flow, fixed-bed reactor described previously (12). The reac-
0
0
tor was constructed of 1/8 -i.d. quartz tubing and contained
a bed of 0.100 g of catalyst. Each experimental run was be-
gun by heating the catalyst from room temperature to 673 K
in the 2.7% thiophene in hydrogen mixture at 0.125 K/s.
The thiophene/hydrogen mixture flowed at 0.5 ml/s over the
bed of catalyst particles with a superficial space velocity of
�
1
1
5 s . Products were identified by gas chromatography and
mass spectrometry (GC/MS) using a Hewlett Packard 5890
gas chromatograph connected in series to a Finnigan Mat
ITD mass spectrometer. Reaction products were quantified
FIG. 2. Thermogravimetric analysis of temperature-programmed re-
action of MoO3 powder and macrocrystals in 85% H2 in nitrogen. Total
using the above-mentioned gas chromatograph equipped mass change corresponds to conversion of MoO3 to Mo2N.

1
40
MARKEL ET AL.
TABLE 1
Summary of Particle Sizes Calculated from X-ray Diffraction
Peak WidthsUsingtheScherrer Equation for Moand Mo2N Crystal
Products
Gas
composition
Temp. ramping D [111] D [200] Sg [111] Sg [200]
2
2
rate (K/min)
(nm)
(nm)
(m /g)
(m /g)
75% H2
75% H2
75% H2
75% H2
75% H2
75% H2
75% H2
0.2
0.6
1
2
5
7.4
8.4
9.6
14.3
17.9
—
6.3
6.9
6.9
7.6
8.6
85
76
66
44
35
101
92
92
82
73
34
10
20
18.2
—
1
2
4
6
8
0% H2
0% H2
0% H2
0% H2
5% H2
0.6
0.6
0.6
0.6
0.6
0.6
13.6
13.0
12.3
11.4
9.6
9.8
9.0
8.0
7.8
6.9
—
46
49
51
55
66
—
64
70
79
81
92
—
1
00% H2
—
FIG. 3. Effect of temperature ramping rate on crystalline MoO3 re-
duction/nitridation.
1
00% NH3
0.6
5.6
4.6
113
137
Macrocrystal characterization. Scanning electron mi-
experiments differed from that observed for air exposed
passivated powder Mo2N catalysts, both of which did not
indicate the presence of MoO2 in the bulk catalyst phase
by means of XRD analysis. This result agrees with previ-
ous reports that passivation gas composition can signifi-
cantly influence the structural properties for the synthesis
of Mo2N unsupported catalysts (24). Because the sample
mass is continuously changing throughout reaction, the
solid transforms through a continuum of states either in-
volving mixtures of stoichiometric compounds or involving
nonuniform, defect (nonstoichiometric) compositions (3).
At higher temperature ramping rates (Fig. 3), reaction
occurs at successively higher temperatures using analogous
temperature programs for each MoO3 macrocrystal sample.
In most cases, the final product of reaction is Mo2N, with
some small amounts of residual MoO3 and Mo observed at
higher ramping rates, as identified by XRD analysis. Pre-
vious work with Mo2N powders reported that the highest
surface areas are produced in syntheses with low reduction
temperatures (25). We were unable to accurately measure
the surface areas of the material produced in the TGA due
to the small sample sizes available but X-ray linewidth data
point to a steady drop in product surface area with increas-
ing temperature ramping rate (Table 1).
croscopy (Fig. 5) reveals the layered platelet habit of the
Mo2N macrocrystals. This layering is seen in the gross mor-
phology as well as in the terraced features at the edges of the
macrocrystals and in samples which have fractured, cleav-
ing along the principal crystallographic planes to yield thin
flakes. The layered habit is an artifact of the MoO3 starting
material. Mo2N does not have an intrinsic layer structure
because it is symmetric along all three crystallographic axes.
Small flat platelets or flakes approximately the size of
typical powdered MoO3 crystallites (<1 mm thick) are ob-
served at the surface of some Mo2N macrocrystals (Fig. 6).
The platelets possibly form when reduced material forms
at the edge of a relatively thin (<10 mm) macrocrystal and
A range of gas compositions were also tested (Fig. 4).
Lower reduction temperatures are observed at the higher
hydrogen contents, as is observed in the powder system
(
22). In pure hydrogen, MoO3 reduces at a relatively low
temperature to form molybdenum metal. In all other cases,
the product of reaction as determined by TGA mass change
and X-ray diffraction is � -Mo2N. Analysis of Mo2N X-ray
linewidths indicates that low temperatures yield higher sur-
face area material (Table 1).
FIG. 4. Effect of gas composition on crystalline MoO3 reduction/ni-
tridation.

HYDRODESULFURIZATION OF THIOPHENE OVER � -Mo2N CATALYSTS
141
FIG. 5. Low-resolution electron micrograph of Mo2N macrocrystal. Scale is indicated by dashed line in title bar.
is sloughed off as reduction progresses into the material.
Powder X-ray diffraction analysis of ground macrocrys-
Reduction of the macrocrystalline MoO3 could begin at tals indicates the product of temperature programmed re-
the edge and progress inward, reducing the oxide to form a action to be � -Mo2N with traces of MoO2 also present (20).
�
more dense phase. This could explain the thinness of the flat The [200] peak of Mo2N at 43.45 is more intense than the
�
product compared to the layered zone which is consumed. [111] peak at 37.5 as has been previously reported for
After the reaction front had penetrated far enough into the high surface area Mo2N powders produced by tempera-
macrocrystal, this phenomena could possibly allow for the ture programmed reaction (12). Scherrer analyses of Mo2N
product material to break away from the parent solid as a diffraction peak widths indicate that the solid diffracting
small platelet or flake.
domains in the ground material average 4 or 5 nm in

1
42
MARKEL ET AL.
FIG. 6. High-resolution electron micrograph of Mo2N flakes on macrocrystal surface. Scale is indicated by dashed line in title bar. Side view of
two partially converted MoO3 crystallites showing consumption of the starting material to form thin product layers which break off to form flakes.
diameter, depending on the lattice plane analyzed. Calcu- are like the much larger bulk crystals and flakes: each is
lated particle diameters do not match because the Scherrer thinner in the [200] direction than in the [111] direction.
analysis assumes that the sample is composed of spheri-
X-ray diffraction scans of oriented Mo2N macrocrystals
cal crystalline Mo2N. In practice, the sample may consist of are shown in Fig. 7. Ifthe smallsolid domainswhich make up
nonspherical polycrystalline particles with a partially amor- the crystal were randomly oriented, these patterns would be
phous surface phase. It is notable that the crystalline dimen- identical to the powder pattern obtained from the ground
sions perpendicular to the [200] axis are consistently shorter macrocrystal sample. However, the diffraction patterns
than the [111] direction. This result was observed previously exhibit only one strong peak corresponding to the (200),
in Mo2N powders. For this reason, the morphology of the (020), and (002) planes. This result would indicate that
nanocrystallites which make up the Mo2N macrocrystals the Mo2N macrocrystals consist of small solid diffracting

HYDRODESULFURIZATION OF THIOPHENE OVER � -Mo2N CATALYSTS
143
domains 4–5 nm in diameter in crystallographic alignment.
Previous work by Volpe and Boudart has examined the
synthesis of Mo2N from the reaction of NH3 with MoO3
powders and crystalline platelets and has determined that
the temperature programmed reaction is topotactic (1).
Topotactic reactions are defined to occur when a solid prod-
uct has well-defined crystallographically orientation rela-
tive to the starting (or parent) solid. The MoO2 observed
in the ground sample has a low specific surface area, as ev-
idenced by its narrow diffraction peaks, and integration of
the observed peaks indicates that MoO2 is present in only
�
trace amounts (20). Only a single peak at 13 is observed
from macrocrystalline Mo2N, instead of the series of peaks
expected for a random arrangement of particles.
Scanning tunneling microscopy (STM) was employed to
investigate both the structure of the solid domains as well
as the pore structure of these materials in Fig. 8. STM of
the [200] face of the topotactic macrocrystal synthesized
FIG. 7. One-dimensional X-ray diffraction patterns produced by ori- by reaction of MoO3 and NH3 shows that the product con-
ented macrocrystals of Mo2N as well as the same macrocrystal ground to
a powder.
sists of rounded particle sizes calculated from X-ray diffrac-
tion peak widths. The measured surface area (44 m /g) of
2
FIG. 8. Scanning tunneling micrograph of Mo2N macrocrystal (200) plane from [5].

1
44
MARKEL ET AL.
the macrocrystal points to an average particle diameter of were identified by GC/MS as C4 hydrocarbons: butadiene,
3 nm. Based on diffraction data, STM measurements, and 1-butene, cis-2-butene, trans-2-butene, and butane. Smaller
1
surface area measurements, it is concluded that the parti- amounts of C2 products were also observed. Product selec-
�
cles are found with a range of diameters from 4 to 13 nm, tivities at 400 C throughout the 12-h reaction period as well
flattened in the [200] direction with either an amorphous as selectivities measured at high space velocities are plotted
surface phase or polycrystalline interior to account for the as a function of fractional conversion of thiophene (Fig. 9).
small Scherrer diameters observed. It was not possible to Fractional selectivities for butane, butadiene, and butene
achieve atomic resolution using these samples, although are strong functions of the thiophene fractional conversion.
such imaging would be useful for verifying the crystallo-
graphic alignments of individual particles.
In all cases, the observed principle product (>55% ) of
desulfurization is butene. Smaller amounts of butane (10–
4
5% ) and even lesser quantities (<3.5% ) of butadiene
Thiophene desulfurization kinetic measurements. Thio- are also observed. Interestingly, the dominant product ex-
phene desulfurization activity over 0.1 g of � -Mo2N macro- pected at equilibrium is butane (26, 27), but only a fraction
�
crystalswasmeasured at 400 C over a 12-h period. The cata- of the product is observed as butane in these kinetic exper-
lyst deactivates 30% during approximately the first 4 h of iments. The observed product distributions are, therefore,
reaction but maintains a relatively constant level of activity due to reaction kinetics.
after that time of approximately 36.1% thiophene conver-
At higher thiophene conversions, the isomerization
sion. The principle products of thiophene desulfurization and hydrogenation reactions appear to be nearly at
FIG. 9. Thiophene desulfurization product selectivities. (a) Butenes, (b) butane, (c) butadiene, and (d) 1-butene/2-butene ratio.

HYDRODESULFURIZATION OF THIOPHENE OVER � -Mo2N CATALYSTS
145
equilibrium with the exception of the butene hydrogena- temperature ramping rate is required for reduction to com-
tion reaction. The butenes are present in thermodynamic mence at low temperature. This result is observed in both
equilibrium amounts (a cis/trans-2-butene ratio of 0.702 and the crystalline and the powder systems.
a 1-butene/2-butene ratio of 0.35) and very little butadiene
Scanning electron microscopy images from the work
is present, as predicted by thermodynamics. However, the of Volpe and Boudart have shown that the reduction of
majority of the products are butenes over the range of con- platelets of MoO3 powder less than 10 �m thick commences
versions observed in this work and these are only slowly at the platelet edges, consuming a portion of the layered ox-
hydrogenated to butane.
ide and forming a reduced product which has the same thin
At differential conversions, 1-butene is the predominant layer morphology but is considerably thinner than the zone
product of thiophene desulfurization. Extrapolation of se- of the oxide which is consumed (1). After a period of reac-
lectivities to zero conversion indicates the direct product of tion, the product material appears to break away from the
thiophene desulfurization is 1-butene with lesser amounts oxide to form product flakes. It is possible that similar mor-
of2-butenesformed and lessthan 10% ofreacted thiophene phology changes occur during the reduction of the macro-
converted to butane. The amount of butadiene produced di- crystals, except the product was not observed to separate
rectly from thiophene is not known because its selectivity from the initialMoO3 macrocrystals. Thisroute isconsistent
curve is asymptotic to the vertical axis, but it is clear that a with the gross observation of macrocrystal oxide samples
significant amount of butadiene is produced directly from darkening first occurring at the edge of the crystal. Similar
thiophene. Changes in selectivities as the fractional thio- phenomena have previously been observed by Volpe and
phene conversion increases are not due to the kinetics of Boudart for the treatment of MoO3 platelets in NH3 below
the thiophene desulfurization reaction, but rather the kinet- 620 K (1). It is possible that the product layers are stabi-
ics of the concurrent C4 hydrogenation and isomerization lized or held in place by layers both above and below in
reactions. Even at differential conversions, the cis and trans single crystal reactions so that individual product layers do
isomers of 2-butene are in equilibrium with cis/trans ratio not readily break off as flakes. In this way, the product of
of 0.702.
reaction consists of a mat of loosely bound Mo2N layers
less than 200 nm thick. Each layer is porous with most of
the observed surface area located in these pores, not the
exposed basal planes of each layer. This porous mat can be
DISCUSSION
Proposed structures of high surface area topotactic peeled easily to expose interior surfaces for analysis.
metallic materials have ranged from highly ordered crys-
High hydrogen concentrations are required to reduce the
talline structures (similar to zeolites, consistent with the crystals at low temperatures and generate surface area. It is
ordered structural changes expected from topotactic re- not clear why relatively modest changes in hydrogen con-
action) to disordered structures (5). In addition, three- centrations(e.g., droppingfrom 75to 30% ) affect the reduc-
dimensional pore structures or two-dimensional, layered tion reaction strongly. Equilibrium effects are not respon-
pore structures are possible. Previous efforts to investigate sible for the observed effects of concentration on reaction:
the poresand solid structure oftopotacticMo2N powdersby for all gas phase compositions used in this study, the equi-
electron microscopies and probe microscopies (e.g., STM librium solid phase is MoO , Mo N, or Mo. No MoO is
2
2
3
and atomic force microscopies) did not produce acceptable expected at equilibrium. Thus, hydrogen concentration has
images of sample topography. Scanning tunneling micro- a strong effect on reaction kinetics, perhaps due to compet-
scopic images of topotactic macrocrystalline Mo2N show itive adsorption effects. It has been shown that H O has a
2
the topotactic material is an aggregate of rounded particles strong effect on reduction kinetics in the powder system at
(
Fig. 8). In addition, X-ray diffraction data show the parti- concentrations as low as 1000 ppm due to competitive ad-
cles are in crystallographic alignment (Fig. 3). This suggests sorption effects (7). High gas space velocities are required
that the particles are interconnected into locked positions in these syntheses because the H O produced by reaction
2
relative to one another to form porous macrocrystals of must be rapidly purged from the bed.
�
-Mo2N.
The specific thiophene desulfurization activity of the
The topotactic reaction kinetics of crystalline Mo2N syn- crystalline Mo N is 2.5 times higher than the activity of the
2
thesis are unlike that for the synthesis of Mo2N powder. powder (Table 2). The origin of this higher activity is not
TGA data from this work indicate the rate of the reduc- known but has been observed previously by this group for
tion reaction is slower for larger crystals than a fine pow- studyofthiophene desulfurization byMo N in powder form
2
der, consistent with a solid-state diffusion limited reaction. with varying surface areas (12). This augmentation in activ-
Previous work with powders has shown that the activation ity for the low surface area nitrided molybdenum powdered
energy of the synthesis of Mo2N from MoO3 is comparable catalysts was attributed to the presence of MoN, which was
with systems involving the solid-state diffusion of oxygen detected using X-ray diffractometry. This result suggested
in metal oxides (28). TGA data also show that a very slow that differences in oxidation state between MoN (+3) and

1
46
MARKEL ET AL.
TABLE 2
The predominance of 1-butene as a direct product of
thiophene desulfurization has been observed in a num-
ber of other systems (12, 30–33). Although the overall ob-
served butadiene content was observed to be less than 1%
at low thiophene conversions (<10% ), the upper limit of
butadiene selectivity was not measured in this work and
could have possibly affected the observed product fractions
at very low thiophene conversions (<1% ). Formation of
butene from thiophene requires a desulfurization mecha-
nism in which four hydrogen atoms are transferred to the
carbon backbone before elimination of sulfur. Most pro-
posed mechanisms involve a sequence of two hydrogenol-
ysis steps and two hydrogenation steps in any of a number
of combinations. An analysis of the possible combinations
of these four steps shows that many are inconsistent with
the observation of 1-butene as the predominant product.
For example, two successive hydrogenolysis steps will lead
to butadiene as the direct product of reaction. Other com-
binations lead to high-energy intermediates or high-energy
products which are expected to be extremely reactive and
should not be observed in significant quantity in the gas
phase or the surface. Two combinations which would lead
to the formation of 1-butene involve (i) two hydrogena-
tion steps to form a 2,3-dihydrothiophene followed by two
hydrogenolysis steps and (ii) a hydrogenolysis step fol-
lowed by a double hydrogenation of the dangling C4 chain
and eventual cleavage to form free 1-butene. Combina-
tion (i) is considered less likely because work with dihy-
drothiophenes has shown that 2,3-dihydrothiophene and
Summary of Measured Thiophene Desulfurization Activities
�
at 400 C and Activities of Reference Compounds
Surface area
Thiophene
conversion (% )
Activity
(mol/s m )
2
2
Compound
(m /g)
Powder Mo2N (ref. 12)
Macrocrystalline Mo2N
MoS2 (ref. 27)
Co0.25MoS (ref. 27)
Co1.5Mo6S8 (ref. 27)
108
44
3.4
10.83
0.15
38.1
36.1
0.76
0.77
0.54
1.95
4.54
0.92
2.92
0.82
Mo2N (+2.5) could in some way influence the activity of
molybdenum nitride catalysts. Daage and Chianelli have
also reported catalyst structure effects for the hydrogena-
tion and hydrodesulfurization of dibenzothiophene (DBT)
over Mo2S powders and microcrystals (14). The work of
Daage and Chianelli reported that both hydrogenation and
hydrodesulfurization were catalyzed over Mo2S powders,
while only the HDS of dibenzothiophene was catalyzed
over Mo2S microcrystals. Daage and Chianelli explained
these results using a “rim-edge” model which attributed the
microcrystalline Mo2S catalyst selectivity for the HDS re-
action to the ratio of exposed basal planes to covered basal
planes for layered Mo2S catalyst particles. Sites located
at the edges of the exposed layers were named rim sites
and those at the edges of interposing catalyst particle lay-
ers were called edge sites. These Mo2S rim/edge site ratios
were correlated to the optical defect density of each cata-
lyst using photodeflection spectroscopy (PDS). Thompson
and coworkers observed similar trends in pyridine HDN
reactions and found that activity normalized more readily
using oxygen uptake values for unsupported catalysts with
compositions similar to those of working catalysts (29).
The observed trends in C4 selectivities are similar to those
observed previously for Mo2N powders (12). Based on se-
lectivity trends at low conversions, it is possible to identify
those C4 products produced directly from thiophene and
trends at higher conversions can be used to qualitatively un-
derstand hydrogenation and isomerization reactions which
accompany the desired desulfurization reaction (Fig. 10).
Desulfurization of thiophene produces primarily 1-butene
along with lesser amounts of 2-butenes, butane, and bu-
tadiene. Before 20% thiophene conversion is reached, the
butene isomers are converted to an equilibrium distribution
of isomers (and cis- and trans-2-butene isomers remain in
equilibrium ratios at all levels of thiophene conversion).
Butadiene reacts rapidly with the vast excess of hydrogen
and is found only at relatively low fractions below 5% at
moderate thiophene conversions. Butene is hydrogenated
relatively slowly to form butane. If allowed to proceed to
equilibrium, all products would be converted to butane ex-
cept for traces of butenes and even smaller amounts of bu-
tadiene (27).
2
,5-dihydrothiophene rapidly isomerize under these reac-
tion conditions, leading to the formation of cis-2-butene
in excess of equilibrium, which was not observed (34, 35).
For this reason, the most likely path for desulfurization in-
volves hydrogenolysis of a single C–S bond, followed by a
dual hydrogenation of the dangling diene chain, followed
by hydrogenolysis of the remaining C–S bond to yield pre-
dominantly 1-butene.
FIG. 10. Proposed thiophene desulfurization reaction pathway
(Mo2N macrocrystal).

HYDRODESULFURIZATION OF THIOPHENE OVER � -Mo2N CATALYSTS
147
CONCLUSIONS
port of this work. We also thank Professor Michael Myrick of the Depart-
ment of Chemistry at the University of South Carolina for assistance with
the scanning tunneling microscopic analyses and Dr. Dana Dunkelberger
of the Electron Microscopy Center at the University of South Carolina
for assistance with the scanning and transmission electron microscopic
analyses.
Macrocrystalline Mo2N has been synthesized from MoO3
macrocrystals by temperature programmed reaction in
N2/H2 and NH3 gases. In this work, Mo2N macrocrystals
2
were revealed to have a specific surface area of 44 m /g.
�
-Mo2N macrocrystals are relatively small, dark metallic-
REFERENCES
gray platelets of moderately porous material which have
visible dimensions varying from micrometer- to centimeter-
scaled proportions.
1
2
. Volpe, L., and Boudart, M., J. Solid State Chem. 59, 332 (1985).
. Ranhorta, G. S., Haddix, G. W., Bell, A. T., and Reimer, J. A., J. Catal.
108, 24 (1987).
. Choi, J.-G., Curl, R. L., and Thompson, L. T., J. Catal. 146, 218 (1994).
. Wise, R. S., and Markel, E. J., J. Catal. 145, 335 (1994).
. Roberts, K. L., and Markel, E. J., J. Phys. Chem. 98, 4083 (1994).
. Volpe, L., and Boudart, M., J. Phys. Chem. 90, 4878 (1986).
. Wise, R. S., and Markel, E. J., J. Catal. 145, 344 (1994)
8. Ranhorta, G. S., Bell, A. T., and Reimer, J. A., J. Catal. 108, 40 (1987).
. Oyama, S. T., Catal. Today 15, 279 (1992).
0. Schlatter, J. C., Oyama, S. T., Metcalfe, J. E., and Lambert, J. M., Ind.
Eng. Chem. Res. 27, 1648 (1988).
From scanning electron microscopic analyses, the synthe-
sis reaction appears to commence at the edges of the pla-
nar MoO3 macrocrystals and later convert the crystalline
MoO3 into a mat of layers each less than 200 nm thick.
However, the bulk of the porosity of the Mo2N macrocrys-
tals does not appear to be located in these exposed basal
planes, but instead in a fine pore structure. Scanning tunnel-
ing microscopy and X-ray diffraction shows that the solid
consists of Mo2N crystallites which are foreshortened in the
3
4
5
6
7
9
1
1
1. Choi, J.-G., Brenner, J. R., Colling, C. W., Demczyk, B. G., Dunning,
J. L., and Thompson, L. T., Catal. Today 15, 201 (1992).
[200] direction with either an amorphous surface phase or a
polycrystalline core. X-ray diffraction patterns from Mo2N 12. Markel, E. J., and Van Zee, J. W., J. Catal. 126, 643 (1990).
1
1
1
1
1
3. Nagai, M., Miyao, J., and Tsuboi, T., Catal. Lett. 18, 9 (1993).
4. Daage, M., and Chianelli, R. R., J. Catal. 149, 414 (1994).
5. Bertrand, O., and Dufour, L. C., Phys. Status Solidi A 60, 507 (1980).
6. Anderson, J., J. Physique, Suppl. 38, C7–C17 (1977).
7. Hillis, M. R., Kemball, C., and Roberts, M. W., Trans. Faraday Soc. 62,
3570 (1966).
macrocrystals indicate the presence of crystallites which are
locked in crystallographic alignment. TGA analyses of the
reduction/nitridation reaction indicate that the reaction is
slower in the macrocrystalline than equivalent powder syn-
theses. The rate of reaction is likely limited by solid-state
diffusion where the reduction reaction would possibly com-
mence at the edges of the MoO3 macrocrystal and consume
the oxide from the edges inward.
18. King, E. G., Weller, W. W., and Christensen, A. U., U.S. Bur. Mines,
Report, 1960.
1
2
2
9. Bursill, L. A., Proc. R. Soc. A 311, 267 (1969).
0. JCPDS X-ray Powder Diffraction Inorganic Reference Data, 1987.
1. Gusev, A. I., Phys. Stat. Sol. (B) 163, 17 (1991).
The thiophene desulfurization activity of macrocrys-
talline Mo2N is observed to be more than twice that of 22. Wise, R., and Markel, E., unpublished results.
2
3. Klug, H. P., and Alexander, L. E., in “X-Ray Diffraction Procedures,”
p. 511. Wiley, New York, 1962.
Mo2N powder on an area basis but is of equal activity on
a mass basis. The reason for this difference in area specific
activities is unknown. Product selectivities for butenes, bu-
tane, and butadiene for the thiophene desulfurization over
24. Colling, C. W., Choi, J.-G., and Thompson, L. T., J. Catal. 160, 35 (1996).
25. Burdick, S. E., and Markel, E. J., unpublished results.
26. Benson, S. W., and Bose, A. W., J. Am. Chem. Soc. 85, 1385 (1963).
Mo2N macrocrystal catalysts are in close agreement with 27. Chase, M. W., Davies, C. A., Downey, J. R., Frurip, D. J., McDonald,
R. A., and Syverud, A. N., JANAF Thermochemical Tables, 3rd ed.
8. Oyama, S. T., Schlatter, J. C., Metcalfe, J. E., and Lambert, J. M., Ind.
Eng. Chem. Res. 27, 1639 (1988).
9. Choi, J.-G., Brenner, J. R., Colling, C. W., Demczyk, B., Dunning,
J. L., and Thompson, L. T., Catal. Today 15, 201 (1992).
those observed for Mo2N powder catalysts. Thiophene is
desulfurized to form predominantly 1-butene with smaller
amounts of other C4 compounds observed.
2
2
ACKNOWLEDGMENTS
30. McCarty, K. F., Anderegg, J. W., and Schrader, G. L., J. Catal. 93, 375
1985).
(
This work was supported by the National Science Foundation Grant
OFR-9108772-02. The National Science Foundation also provided fund-
31. Osbaldiston, R. J. C., MS thesis, Univ. of South Carolina, 1993.
32. Desikan, P., and Amberg, C. H., Can. J. Chem. 41, 1966 (1963).
ing for the electron microscope (Equipment Grant DIR-9016370). K. L. 33. Kolboe, S., and Amberg, C. H., Can. J. Chem. 44, 2623 (1966).
Roberts also thanks the Department of Energy EPSCoR Fellowship pro- 34. Markel, E. J., Sauer, N. N., Angelici, R. J., and Schrader, G. L., J. Catal.
gram at the University of South Carolina and the Department of En-
ergy Samuel Massie Chair of Excellence in Environmental Engineering at
North Carolina Agricultural and Technical State University for their sup-
116, 11 (1989).
35. Sauer, N. N., Markel, E. J., Schrader, G. L., and Angelici, R. J., J. Catal.
117, 295 (1989).