CARBON MOLECULAR SIEVES AS CATALYSTS,
CATALYST SUPPORTS, AND IN ADVANCED MATERIALS
GERALD C. GRUNEWALD
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
TO MY FAMILY FOR THEIR CONSTANT SUPPORT
There have been a large number of people who have helped to make
my journey through graduate school very enjoyable. The list presented
here is by no means complete. Foremost, I would like to express my
gratitude to my research advisor, Russell Drago, whose support and
guidance have been essential to my education. Besides showing me the
excitement of chemical research, Doc has also contributed significantly
to my knowledge of two-on-two basketball and has taught me the fine art
of eating spaghetti without cutting it. I would equally like to thank
Ruth Drago whose humor and hospitality helped to make Gainesville a home
away from home. Maribel Lisk also deserves special thanks. Her
friendship and assistance, along with her seemingly endless supply of
stationery goods, have been most appreciated.
My coworkers during my years in the Drago group have been a
valuable source of knowledge and advice as well as being great friends.
Former group members Jeff Clark, Andy Griffis, Ken Balkus, Rich Riley,
Pete Doan, Ed and Cindy Getty, and Shannon Davis helped a great deal in
showing me "the ropes." Present group members Larry Chamusco, Ngai
Wong, Bobby Taylor, Tom Cundari, Al Goldstein, Mike Naughton, and Steve
Petrosius have been comrades, compatriots, and cohorts in both
scientific and social endeavors. I would also like to thank my
roommates of the renowned Thunderdome. Mark Hail, Mark Barnes, Don
Ferris, and Steve Showalter have been great friends as well as often
being good cooks and housecleaners. A special thanks is also due to the
Stranger band who provided many a night of great entertainment for the
revelers of the Inorganic and Analytical Divisions.
One other former Drago group member has also had a significant
influence on my career. I would like to thank my future wife, Cindy
Bailey, for her love, support, and advice. Although there are a lot of
things I will miss in Gainesville, the frustration of a long-distance
relationship and the high phone bills will not be among them. Lastly, I
would like to thank my family for their constant prayers and support.
In particular, I want to thank my brother-in-law, George Fisher, who was
a great help in preparing this manuscript.
TABLE OF CONTENTS
LIST OF TABLES . .
LIST OF FIGURES .
ABSTRACT . .
GENERAL INTRODUCTION .
PREPARATION AND CHARACTERIZE
3 CARBON MOLECULAR SIEVES
Results and Discussion
4 CARBON MOLECULAR SIEVES
Results and Discussion
5 CARBON MOLECULAR SIEVES
Results and Discussion
6 CONCLUSIONS .
REFERENCES . .
BIOGRAPHICAL SKETCH .
TION OF CMS MATERIALS
. . .
CATALYST SUPPORTS .
. . .
. . .
. . .
. . .
LIST OF TABLES
Elemental Analysis of PAN Pyrolysis in
Various Carrier Gases . .
Surface Area of PPAN Samples .
Physical Properties of AX21 . .
Elemental Analysis of AX21 .
Ethyl Benzene Reactivity Over PPAN
Catalysts at 2500 C . .
Ethyl Benzene Reactivity Over PPAN
Catalysts at 3500 C . .
Ethyl Benzene Reactivity Over AX21
and Other Catalysts at 3500 C .
Butene Reactivity Over CMS Catalysts .
Ethanol Reactivity Over CMS Catalysts .
Reactivity of C3 Oxygenated Substrates
Over AX21 Catalysts . .
Commonly Studied Fischer Tropsch Systems
Typical Methanol Oxidation Reaction Routes
Effect of Carrier Gas Cycling on Acetaldehyde
Reactivity Over Ru/PPAN . .
Methanol Reactivity Over Molybdenum Doped Catalysts
Deep Oxidation of CH2CL2 Over Doped AX21 Catalysts
Results of LiC1/PPPO Ion Selective
Electrode Experiments . .
LIST OF FIGURES
Formation and reaction cycle of PPAN .
Formation of PPPO . .
Pyrolysis profile of PPAN . .
Formation of pyridone moieties in PPAN structure
DRIFT spectra of PAN and PPAN . .
Infrared comparison of PPAn samples prepared
under various conditions . .
Infrared spectrum of PPO before pyrolysis .
Infrared spectra of PPO during pyrolysis .
DSC of undoped PPO . .
DSC of Ti(O) doped PPO . .
DSC of LiCl doped PPO . .
DSC of CoC12 doped PPO . .
Schematic diagram of reactor setup .
Proposed mechanism for AX21
dehydrogenation activity . .
Proposed mechanistic routes for
conversion of n-propanol . .
DRIFT spectra of PPAN and Ru3(CO)12/PPAN ..
DRIFT spectrum of unsupported Ru3(CO)12 .
DRIFT spectrum of Ru3(CO)12 . .
GC traces for reactant gas cycling experiments
Mass spe crum of propane peak
showing C incorporation . .
Product distribution of Ru/PPAN versus Ru/A1203
Alkene production in Ru/PPAN system .
ASF plot of PPAN versus A203 .. .
SEM of Ru/A1203 before and after catalysis .
SEM of Ru/PPAN before and after catalysis .
Product distribution of Ru/AX21 versus Ru/A1203
ASF plot of AX21 versus A1203 .. ...
Effect of CO addition to CO2 methanation reaction
Effect of acetaldehyde addition to CO/H2 reaction
Proposed mechanism for product enhanceme
from aldehyde addition .
on ... 120
on ... 121
on ... 122
SEM of bulk Mo03 before and after
SEM of Mo03/SiO2 before and after
SEM of MoO3/AX21 before and after
XRD of Mo03/SiO2 before reaction
XRD of Mo03/AX21 before reaction
XRD of Mo03/SiO2 after reaction
XRD of Mo03/AX21 after reaction
XPS of the Mo 3d orbitals of bulk
XPS of the Mo 3d orbitals of
Mo03/AX21 pre-catalyst .
XPS of the Mo 3d orbitals of
Mo03/SiO2 pre-catalyst .
XPS of the Si 2s orbital of
MoO3/SiO2 pre-catalyst .
XPS of the Mo 3d orbitals of
Mo03/AX21 post-catalyst .
XPS of the Mo 3d orbitals of
Mo03/SiO2 post-catalyst .
4-29. XPS of the C Is orbital of
MoO3/SiO2 post-catalyst .... 138
5-1. Schematic of cell design in ion
selective electrode experiments .... .153
5-2. Cyclic voltammogram of Ti/PPPO
system in N2 and 02 . . 156
5-3. Effect of pH on Ti/PPPO 02 reduction system .... 157
5-4. Effect of cathodic and anodic extreme potential
variation on Ti/PPPO 02 reduction system. ... 159
5-5. Effect of scan rate variation on
Ti/PPPO 02 reduction system . .. ... 162
Abstract of Dissertation Presented to the Graduate School of
the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
CARBON MOLECULAR SIEVES AS CATALYSTS,
CATALYST SUPPORTS, AND IN ADVANCED MATERIALS
Gerald C. Grunewald
Chairman: Russell S. Drago
Major Department: Chemistry
The ability of activated carbons prepared from both natural and
synthetic precursors to function as molecular adsorbents has been known
for a very long time. Such materials operate efficiently as general
adsorbents because their pore sizes are very large and thus they are
unselective. In addition to their excellent adsorption capabilities,
carbonaceous materials have also been reported to catalyze
dehydrogenation reactions and a great deal of work has been done with
carbon based systems in advanced materials applications. For the most
part, however, typical charcoal and other activated carbon adsorbents
differ widely from one sample preparation to the next, are commonly
poorly characterized, and generally are not considered to be chemically
In this study, Carbon Molecular Sieves (CMS) materials prepared
from the pyrolysis of polymeric precursors have been investigated as
catalysts, catalyst supports, and in advanced materials applications.
These CMS systems are shown to be active catalysts for the oxidative
dehydrogenation and dehydration of a variety of substrates. In several
cases, the activity of these materials exceeds that of inorganic oxide
based catalysts. Metal doped CMS systems have also been studied in
several different types of reactions and the CMS supports are found to
have inherent advantages over the widely employed inorganic oxides.
Synergistic interactions between dopant metal species and CMS supports
are observed and characterized. The CMS systems have been further
investigated for use as novel electrodes, electrocatalysts, and
Inorganic oxides such as alumina and silica have been widely
studied and employed in a host of industrial processes as heterogeneous
catalysts and catalyst supports for the past fifty years.1-3 The
introduction of synthetic zeolites in the early 1960s sparked a further
surge of research into materials science and catalysis.4 In the ensuing
years hundreds of patents, books, and papers have described the
application of zeolites to a plethora of separation processes and shape
selective catalysis systems. Currently a new class of materials has
been prepared which has the potential to equal or surpass the alumino-
silicate compounds in industrial operations. Carbon Molecular Sieves
(CMSs), which are produced from the controlled pyrolysis of polymers or
polymeric precursors, are being actively studied. Initial
investigations have shown these materials to be very effective in gas
separation studies,5-7 pressure swing absorption experiments,8 and in a
variety of catalytic systems.9-11 Despite these encouraging preliminary
results, there have been surprisingly few publications on CMS systems
and the majority of these reports are in the European and Japanese
patent literature.12-17 In discussing CMS's, it is important to note
that the term has been used in the literature to describe a wide variety
of materials. Studies of carbonaceous compounds with sieve-like
properties were reported as early as the late 1930's.1821 Much of this
work focused on natural organic substances such as coals and charcoals
which had sorption properties generally related to their porosity. Due
to their amorphous nature, however, the pore sizes of these materials
ranged a great deal between samples. More versatility and
reproducibility in pore dimensions were obtained through the pyrolysis
of synthetic polymeric precursors. Pyrolyzed polyvinylidenechloride was
determined to have slit shaped pores which were efficient in separating
molecules based on steric limitations.22 Branched hydrocarbons such as
isobutane were successfully separated from their straight chain isomers.
Sieves based on pyrolyzed polyfurfuryl alcohol were shown to have
similar separation capacities. These materials were even employed as
catalytic supports in hydrogenation reactions. In separate studies,
Trimm and Cooper23'24 and Schmitt and Walker25 demonstrated that
platinum doped pyrolyzed polyfurfuryl alcohol would selectively
hydrogenate linear over branched alkenes. The catalytic hydrogenation
was due to the metal, but the sieve provided the selectivity. The most
effective polymeric precursor to date has been polyacrylonitrile.
Sieves composed of pyrolyzed polyacrylonitrile have been established to
have shape selectivities greater than 100:1 for molecules which differ
by as little as 0.2 A in their critical diameter.26 The most recent
development in carbon molecular sieve manufacturing has been the thermal
activation of petroleum coke and coal tar pitch residues to produce
materials with very high surface areas.27,28
The research described in this dissertation concentrates on three
different CMS systems for use as catalysts, catalyst supports, and in
advanced materials applications. Pyrolyzed polyacrylonitrile (PPAN) has
been prepared, characterized, and studied as a catalyst in oxidative
dehydrogenation reactions and as a support in heterogeneous catalytic
systems for synthesis gas (CO and H2) conversion. AX21, a commercially
available CMS, has been characterized and studied as a catalyst in both
dehydration and dehydrogenation reaction schemes and as a synergistic
support in a variety of catalytic transformations. Pyrolyzed poly (2,6-
dimethylparaphenylene) oxide (PPPO) has been prepared, characterized,
and studied in several advanced materials applications as a novel
support for new electrocatalysts and as ion selective electrodes. The
specific characteristics of these materials which make them ideal for
the studies described are discussed below for each of the three CMS
The interesting structural and physical changes that occur during
the pyrolysis of polyacrylonitrile (PAN) have been studied periodically
for the last thirty years. Burlant and Parsons reported that upon
thermal treatment, PAN undergoes a chemical reaction leading to
discoloration and the appearance of an EPR signal.29 IR studies
attribute these observations to a cyclization and dehydrogenation
process as shown in Figure 1-1.30-37 The formation of this doubly
conjugated ladder structure is believed to give the material its
semiconductor nature.38-40 Due to the ability of the condensed pyridine
ring moiety of PPAN to be hydrogenated (step 4 of Figure 1-1) and
subsequently dehydrogenated via air oxidation (step 5 of Figure 1-1),
PPAN has been studied for a number of catalytic processes. The
dehydration and dehydrogenation of alcohols,41 the isomerization and
dehydrogenation of olefins41,42 and the dehydrogenation of ethyl
H2 H2 H2 H2
-C -C 1 C H C H
C CC C
I I I PAN
C C N C N C
^N N ^N ^N
H1 H2 H12 H2
C H C H CHC.H/
C C C C
N/ .N N '*NN/ N
H H H H
Hz Hz Hz Hz
CHCHICHC 1H /
C C C C
C C C
N/ N N/ N
I I I
H H H
H H H H
.C0 C C C
I (I C + H20
N/ -N N N/C
Figure 1-1. Formation and reaction cycle of PPAN.
benzene43 have all been studied on various PPAN systems. Metal doped
PPAN catalysts have also been employed for reaction schemes such as the
decomposition of nitrous oxide,44 the epoxidation of ethylene'45 and the
oxidation of cumene and ethyl benzene.46
AX21 is a truly remarkable CMS. It is prepared by a patented
process involving the high temperature pyrolysis of polymeric petroleum
coke in the presence of large amounts of KOH.28 The surface area of
this material, as measured by the standard BET method,18 is in excess of
3000 m2/g. This value underscores the extraordinary nature of the
material. The theoretical maximum for surface area based on a monolayer
of carbon including the area on both sides of the plane is 2620 m2/g.47
Exposure of both sides of every carbon atom is an impossibility for a
solid with any physical integrity. Thus this compound is on the
threshold of sensitivity for surface area measurement techniques and is
among the most porous materials ever synthesized. In addition to the
phenomenal surface area, AX21 also has exceptional adsorption
capabilities. Studies have shown that this material has four to five
times the adsorption capacity of other typical highly adsorbent
carbonaceous compounds.4850 These properties make AX21 an attractive
candidate for catalytic studies since high surface area combined with a
strong affinity for a substrate are key parameters for successful
Poly (2,6-dimethylparaphenylene) oxide (PPO) is a very versatile
polymer which is used for a variety of purposes. The high decomposition
temperature, due to the stability of the ether linkage, makes this
material ideal for applications such as heat-exchange fluids and high-
temperature lubricants.51'52 The pyrolysis of this polymer cleaves some
of these ether linkages and causes a significant amount of cross-
linking. This process leads to a product having functionalities such as
those shown in Figure 1-2.53 With respect to industrial lubricants and
other similar applications, the formation of this highly crosslinked,
completely insoluble, material is clearly not desirable. This compound
is, however, potentially useful as a template in which a variety of
dopants can be added and the chemistry of the dopant probed while it is
contained in the pyrolyzed polymer matrix. This type of host-guest
interaction has not been investigated previously for dopants such as
metals or ionic salts; however, some polymer blend studies have shown
that PPO can stabilize other polymers in thermal degradation studies.
Polystyrene (PS) has been shown to be thermally stabilized in the
presence of PPO due to a cage-like effect in which the PPO encapsulates
the PS constituent.54
In studying these three particular CMS systems, the research
centers on the novel chemical structure and reactivity of these
materials which impart to them unique properties. The specific
reactions and applications that are studied in this work are chosen in
order to utilize these exceptional features. The focus of the research
is on the treatment of CMS materials as chemical entities. Much of the
work done to date has been in studying the physical nature, e.g. pore
sizes and adsorption capacity, of CMS systems. The work presented here
shows that the chemical reactivity of various CMS materials is quite
versatile and is potentially applicable to a variety of industrial
L n"3 n
Figure 1-2. Formation of PPPO
Figure 1-2. Formation of PPPO
PREPARATION AND CHARACTERIZATION OF CMS MATERIALS
Pyrolyzed Polyacrylonitrile (PPAN)
Although the condensed pyrdine ring structure of PPAN shown in
Figure 1-1 has been invoked by virtually all researchers who have
studied this material, most agree this is an over simplification of a
much more complex structure.32-37 Upon initial observation, the
formation of the conjugated ladder structure appears straightforward.
The nitrile group acts as a nucleophile to attack the adjacent chain.
This is followed by dehydrogenation of the newly formed ring to yield
the proposed structure. In practice, however, the synthetic path to
PPAN has many side routes. PAN is an atactic polymer. The oligomer
shown in Figure 1-lb is drawn as an isotactic fragment. When present in
this configuration, the "zippering" of the backbone to form the cyclized
product can readily occur. Unfortunately, PAN has not been synthesized
with any large degree of stereo regularity on a bulk scale.55,56 Some
work has been done on the electrocatalytic deposition of thin PAN films
on metal electrodes which is proposed to yield isotactic species,
however, the polymer coating formed is only a few monolayers thick.57,58
Clearly, any hope to use these materials as catalysts requires a
preparative method that can generate reasonably large quantities. Due
to this lack of isotacticity, the bonding rearrangements that occur
during the thermal process can proceed in different directions and give
rise to a significant amount of crosslinking. In light of the
complexity of this synthetic process, it is easy to understand why there
are many variations among PPAN materials that have been reported.
In this research effort, PPAN materials are prepared for use in a
variety of catalytic studies as both catalysts themselves and as
catalyst supports. A number of different substrates as well as a
variety of dopants are examined. Because of the nature of this work,
which involves continuous comparisons to be made from experiment to
experiment to optimize conditions, it is essential that the PPAN
materials be the same from each preparation. Furthermore, since the
proposed active species for dehydrogenation catalysis is the condensed
pyridine ring structure, the preparation employed should maximize the
content of this active species in the final product. The two most
important parameters to control in order to accomplish these goals are
the pyrolysis procedure and the nature of the carrier gas. Among the
reported PPAN studies, a wide variation in heating cycles and carrier
gas compositions have been employed. A lot of time has been spent in
our laboratory in designing optimal conditions for reproducible PPAN
Thermo-Gravimetric Analysis (TGA) and Differential Scanning
Calorimetry (DSC) were carried out by Jeff Clark at the University of
Florida and the results are described in this Ph.D. dissertation.59
This work was used to help design a suitable pyrolysis procedure. In
isothermal TGA studies, weight loss occurs sooner and to a greater
extent as heating temperature is increased in successive experiments.
Temperature programmed TGA work shows similar results. Faster heating
rates cause increased weight loss. The DSC studies compliment the TGA
experiments. The onset of weight loss corresponds to a large exotherm.
In isothermal studies, the sharpness of the exotherm increases with
temperature. Temperature program DSC work shows that the size and
specific temperature of the exotherm are directly proportional to scan
rate. The simultaneous weight loss and exotherm are proposed to be due
to the cyclization of the nitrile groups. Further weight loss and a
much smaller exotherm are due to the last dehydrogenation step. It is
clear from these results that the heating process should be done in
stages and that the progression from step to step should be done very
slowly. If the heating rate is too fast, the exothermicity of the
process will cause a runaway reaction leading to large weight loss,
extensive crosslinking and the formation of a very complicated network.
The pyrolysis profile shown in Figure 2-1 is found to yield
samples which are reproducible in their color homogeneity (black), EPR
signal (very intense organic radical with g = 2.01), elemental analysis,
and Infrared spectra. This heating cycle is different from that
employed in most of the reported studies. Several of the literature
preparations invoke simply heating isothermally at 200-2500 C for a few
hours.38'39,46,57 Another common preparation involves a two-step
process where the sample is initially heated isothermally at 2000 C in
an air atmosphere and then subsequently heated at 4000 C in a N2
atmosphere.31-37 It has been our experience that the first of these
Figure 2-1. Pyrolysis profile of PPAN.
methods does not fully pyrolyze the sample. The color of the product is
usually brownish and there is still a significant amount of nitrile
functionality observed in the IR of the product. The second method does
yield a homogeneous black product with no nitriles remaining; however,
both elemental analysis and IR show that there is a lot of oxygen
incorporation in the final material.
The presence of oxygen in the end product has proven to be a
ubiquitous observation among PPAN samples prepared under a variety of
conditions. Table 2-1 contains the elemental analysis results of PPAN
samples pyrolyzed in several different carrier gas atmospheres. It is
evident that the product is quite sensitive to oxygen incorporation even
under what should be anaerobic conditions. There is a significant
amount of disagreement in the literature over this observation. Grassie
and McGuchan pyrolyze in a N2 atmosphere and still observe the presence
of oxygen.32-34 They invoke a pyridone type structure such as that
shown in Figure 2-2 to account for this. Coleman and Petcavich observe
the presence of oxygen containing species even when the pyrolysis is
carried out under reduced pressure.35 Rafalko as well as Chung and co-
workers, however, report elemental analyses consisting of over 99% C, H,
and N.36,37 In comparing these various literature results to the
present work, again there are important experimental differences to be
noted. All the literature studies cited use thin films of PAN weighing
about 100 mg. In order to produce enough material for a series of
catalytic experiments from the same batch, it is necessary to use 5-7g
of PAN in the pyrolysis step. This large sample size makes it difficult
to remove all intestitial air which is felt to be the source of the
Elemental Analysis of
PAN Pyrolyses in Various Carrier Gases
PYRnI Y~TS ATMOSPHFRF
C % %H %N TOTAL
The non C, H, or N constituent is found to be 0.
,,, F .v...V-... A.I-,PER %C %H %N TOTAL
H H H
H H H H2
Figure 2-2. Formation of pyridone moieties in PPAN structure.
incorporated oxygen. Due to the exothermicity of the structural
transformations, it is not advantageous to tightly pack pyrolysis tubes.
When the samples are packed together, the heat generated from the first
cyclization raises the temperature of the system too quickly and causes
undesirable charring. Thus, the PAN powder is layered out in a
horizontal pyrolysis tube.
The presence of the oxygen would suggest that there is a
significant amount of the pyridone moiety in the product prepared in
this work. Infrared studies, however, do not support this. Figure 2-3
shows the complete spectrum overlay of both PAN and PPAN. The key
comparative features are the loss of the C=N stretch at 2240 cm-1 and
the growth of broad peaks from 1600 1200 cm-1. These can be assigned
to the various C=C, C=N, and C-C, C-N functionalities. Figure 2-4 is an
expansion of the essential double bond region. This spectrum is
displayed in the absorbance mode so that it can be directly compared to
the spectrum reported by Coleman and Petcavich.35 The absence of broad
shoulders in the 1700 cm-1 region, which is indicative of C=0 stretches
resulting from pyridone moieties, suggests that there is not an
appreciable amount of pyridone functionality in the samples prepared by
our method. The oxygen present may be involved in chain terminating
hydroxyl species, since there are broad absorptions above 3200 cm-1
After devising a method of preparing reproducible PPAN samples
with a large amount of condensed pyridine ring functionalities, the next
goal was to increase the surface area of the product. Results of
studies employing PPAN materials as catalysts and catalyst supports are
discussed in the following two chapters. One of the key parameters for
3800. 0 3300.0 2500. 0 2300. 0 1800. 0 1300. 0 800.00
DRIFT spectra of PAN and PPAN.
)300 O 1CO 2 C00 100
1l= iaoo 0oo
3500.0 3100 0 2700.0 2300, 0 I0o I) 1500.0 1 100. 700. 00
2500 4C 2000
Infrared comparison of PPAN samples prepared under
either of these applications is to have as high a surface area as
possible. The pyrolysis of PAN according to the method described here,
as well as the various literature methods, yields a product with a very
Reported surface areas range between 8-17 m2/g.
PPAN samples prepared in this study
the 9-12 m2/g range. In order to in
pyrolysis agents are investigated.
various dopants have on the surface
is evident that the most success is
and NH4C1. NH4C1 decomposes to NH3
procedure. As the gases volatilize
greatly increase the surface area.
occurs at 3600 C, it is an excellent
are found to have surface areas in
crease these values, a series of co-
Table 2-2 outlines the effects
area of the final PPAN product. It
achieved with a 50-50 mixture of PAN
and HC1 during the pyrolysis
out, they create channels which
Since the decomposition of NH4C1
dopant to employ. The final
dehydrogenation step leading to the fused pyridine ring structure is
proposed to occur between 300-3200 C. Thus the volatization of the
dopant occurs after the desired structure is formed. The surface area
of this material is 50 m2/g which is about an order of magnitude higher
than the undoped PPAN. IR and elemental analysis on the product of this
doped sample verify that all of the dopant is removed by the end of the
pyrolysis procedure. Catalytic studies videe infra) show that this
higher surface area increases the reactivity of the sample.
The AX21 carbon molecular sieve is purchased from the Anderson
Development Company. The material is made by a direct chemical
activation route in which petroleum coke is reacted with excess KOH at
low surface area.
Surface Area of PPAN Samples
E CAFRUS ARFA m2/a
none N2 pyrolysis
none air pyrolysis
a) These sample names will be used throughout the rest of this text
when describing particular PPAN material.
b) These pyrolyses are all done in N2. Dopant percentages are
based on weight. The samples are stirred in 300 ml H20 after
pyrolysis to wash out any remaining dopant material.
,Zrll Pa ,_, .
5000 C. This produces an intermediate product that is subsequently
pyrolyzed at 9000 C to yield active carbon which contains potassium
salts. These salts are removed by successive water washings. More
detail on the preparation of this material is reported elsewhere.60
Since the AX21 material is commercially available, most of the
characterization studies have been done by the producer. A list of the
more relevant ones to catalytic studies is given in Table 2-3. As was
mentioned in Chapter 1, the reported surface area of this material is
known to have been overestimated. The theoretical maximum for a
carbonaceous material is 2620 m2/g.47 This inflated value, however,
exemplifies the unusual nature of this material.
Transmission Electron Microscopy (TEM) shows this material to be
composed of interwoven carbonaceous lamellae that form a three
dimensional network with molecular size channels throughout.61 These
channels produce pores which give the material its sieving properties.
Unlike standard activated charcoals, the pores of the AX21 material are
considerably smaller and have a very narrow size range. The pore size
for AX21 is 5-7 A, whereas that for typical bone charcoal can range from
twenty to several hundred A.62 Another difference between the AX21
compound and other typical carbonaceous adsorbents is its incredible
adsorption capacity. Studies have shown this material to be far more
adsorbent than other widely used CMS compounds.4850
Outside of the adsorption properties and pore structure, these
materials have proven to be rather difficult to characterize--
particularly from a chemist's point of view. Even elemental analysis, a
seemingly facile technique, became more difficult than anticipated.
Table 2-3 Physical Properties of AX21
Surface Area, BET, m2/g
Total Pore Volume, ml/g
Bulk Density, g/ml
passes 100 mesh, wt%
passes 200 mesh, wt%
passes 325 mesh, wt%
Table 2-4 lists the elemental composition of a very dry AX21 sample. To
obtain this data, the sample had to be run at least three separate times
for all the elements and a total of five times for the carbon
quantification. The necessity of the subsequent runs was not that the
sample was heterogeneous, but rather that it didn't always combust
completely. Some of the carbon became refractory, i.e. non-combustible,
and thus led to inaccurate results. The 2.5% elemental composition
which is unaccounted for in Table 2-4 is felt to be due mostly to the
carbon not fully combusting, as opposed to the presence of other
impurities not screened for in the analysis. The experimental support
for this contention comes from Electron Diffraction Spectroscopy (EDS)
which was done in conjunction with Scanning Electron Microscopy (SEM) on
several of the metal doped AX21 samples. The results of these studies
will be discussed in more detail in Chapter 4, but it serves to mention
here that no other elements, outside of the metal dopant and those
listed in Table 2-4, were observed in any of the elemental analysis
scans from EDS.
Other common characterization techniques such as IR and EPR do not
reveal much information. Unlike PPAN, which can be studied quite well
with the DRIFT technique, AX21 displays a completely featureless
spectrum. Since elemental analysis shows in excess of 6% oxygen
content, it is surprising that no hydroxyl, carbonyl, ether, or any
other oxygenated chromophore is observed. IR analysis is even more
unusual when doped AX21 samples are attempted. Figure 2-5 shows the
spectrum overlay of Mo03 alone and MoO3/AX21 in transmittance mode. As
is shown, the spectrum for the latter system appears as an inversion of
Table 2-4 Elemental Analysis of AX21
Note: The cumulative error associated with all these separate
analyses causes the theoretical maximum total to be only 99%.
a) Values varied from 88.6 91.2%.
b) Values varied from 1.29 2.39. Composition is virtually all
Infrared spectrum of PPO before pyrolysis.
the dopant. Apparently, the AX21 material absorbs all of the IR energy
via some type of charging mechanism. In all the DRIFT experiments the
sample is mixed with about a hundred fold excess of KBr. Thus, the
total absorption is not due to the opaqueness of the sample. In fact
the reflected beam throughput of the AX21 samples is higher than many of
the PPAN samples. The total absorption of the IR energy is due to the
AX21 material and is not an instrumental artifact.
Since the PPAN samples are highly paramagnetic and AX21 is
prepared from pitch residues, it was expected that the latter material
would also have an intense delocalized organic radical signal in its EPR
spectrum. Initially it was difficult to obtain a spectrum at all
because the samples absorb H20 so readily. When excessively dry,
however, the AX21 material is observed to have no paramagnetic content.
Much characterization work has been done on the metal doped AX21
system. The results of X-ray Diffraction (XRD), X-ray Photoelectron
Spectrocopy (XPS), and Scanning Electron Microscopy (SEM) are discussed
in Chapter 4. For the most part though, these techniques characterize
the dopant and its interaction with the AX21 support more than the AX21
material itself. Several new methods are currently being applied to
these materials which may prove to be quite informative. 129Xe NMR,
which can yield information regarding the chemical nature of the pore
site environment, is just beginning to be investigated by several
research groups.63-67 Although there have not been any studies published
on CMS systems yet, this may prove to be a promising technique in the
future. Titrametric calorimety is another method which can yield
fundamental chemical information on solid materials. Drago and co-
workers have studied a Pd/charcoal system and have been able to quantify
the acidic nature of this material.68 The extension of this technique
to CMS materials such as AX21 could provide a wealth of information.
Pvrolvzed Poly (2,6 dimethylparaphenlvene) oxide (PPPO)
As is shown in Figure 1-2, the pyroloysis of poly (2,6-
dimethylparaphenlyene) oxide (PPO) produces a highly crosslinked matrix.
The major experimental evidence to support this proposed structure comes
from Pyrolysis IR studies. Figure 2-6 shows the IR spectrum of PPO at
three different temperatures during its pyrolysis. It is evident that
there is no real change in the general features of the spectrum as the
material is pyrolyzed. The only noticeable variation is a gradual
decrease in the intensity of all the peaks. Concomitant with this
overall decrease in intensity, is the observation of a color change from
white to yellow to eventually black. This darkening of the sample is
the cause of the decrease in intensity. Unlike PPAN, pyrolysis does not
bring about any new functional moieties, but rather just rearrangements
of the ether linkages to produce a highly interwoven, template
The advanced materials applications discussed in Chapter 5 involve
the entrapment of various metal and ionic species in the cavities of the
PPPO matrix. To verify that the presence of the dopants does not change
the nature of the PPPO structure, Pyrolysis IR studies were done on both
Fe(O)/PPO and LiCI/PPO composites. In both cases, the spectra do not
differ from that of the PPO itself. EPR spectroscopy done on the
undoped material shows a very large delocalized organic radical with a
B C E
2 Hrs. Heating 114 C
1\ I H 857
4 Hrs. Heating
4.5 Hrs. Heating
Infrared spectra of PPO during pyrolysis.
g value of 2.01. This is essentially the same as that for the PPAN
Thermo-Gravimetric Analysis (TGA) and Differential Scanning
Calorimetry (DSC) were conducted on the undoped and several of the doped
composites. The TGA shows virtually no weight loss until 3300 C in any
of the samples. As is discussed in the experimental section of this
chapter, the pyrolyses in this study are done isothermally at 2500 C.
This lack of any weight loss supports the proposal that the sample just
crosslinks and does not produce any significant amount of volatiles.
The DSC results are worthy of further discussion. Figure 2-7 shows the
DSC for the PPO material itself. It is evident that there are
essentially no endo- or exothermic processes occurring in the
temperature range studied. A very small endotherm is detected at
2100 C, but this is on the threshold of the instrumental limitations.
Figure 2-8 displays the DSC for a Ti(0)/PPO sample. Again the spectrum
is virtually nondescript save an even smaller endotherm at 2100 C. Thus
the polymer itself and zero valent metal doped composites undergo no
discernable thermochemistry in the temperature range of interest.
Samples containing doped metal salts, however, behave significantly
different. Figure 2-9 shows the DSC for a LiC1/PPO composite. There is
a reasonably large endotherm at 2470 C. The heat absorbed by the system
corresponds to 10.33 J/g. Figure 2-10 displays the DSC for a CoC12/PPO
composite. Like that of the LiCl doped system, there is an appreciable
endotherm. The peak maximum occurs at 2460 C and has a value of
13.55 J/g. Each of these DSC studies were run several times and were
found to be quite reproducible. Slower, faster, or isothermal heating
DSC of undoped PPO.
DSC of Ti(O) doped PPO.
PEAK FROM: 227.768
170.00 190.00 210.00 230.00 250.00 270.00
290.00 310.00 330.00
DSC of LiCI doped PPO.
DSC of CoCl2 doped PPO.
made little or no difference in the peak height or peak location in any
of the samples. These results suggest that there is some process driven
by entropy occurring with the salt doped samples that is not going on in
the other systems. This is most likely associated with the dissolution
of the ionic dopant in the polymer matrix. The majority of the
endotherm would be a function of the lattice energy of the salt with
some minor contributions possibly coming from structural rearrangements
of the PPO matrix to accommodate ionic species. Since the endotherm is
greater for a divalent ion than a monovalent ion, the is support for the
proposal of lattice energy being an essential component of the process.
Polyacrylontrile samples were purchased from both Aldrich and
Polysciences. Additional samples were prepared by the radical initiated
polymerization of acrylontrile according to literature methods.31 No
significant differences were observed in the PPAN product prepared from
the various PAN sources. The pyrolyses were carried out in a horizontal
tube furnace. Typically 5g of the sample were layered out in a pyrex
tube fitted with a male 24/40 ground glass joint at one end and tapered
to a standard size hose connection spout at the other. Initially a
sintered glass frit was employed at the tapered end, but these were very
difficult to clean and often retained the liquid effluents of the
pyrolysis degradation products. Subsequent runs simply employed a small
glass wool plug at the tapered end. The PAN sample is slowly poured in
and layered out over a 3-4 inch space. This sample is usually not in
contact with the glass wool plug. As mentioned previously, tightly
packing the sample into the tube is not advantageous because the
exothermicity of the process causes packed sample to char. Carrier flow
(10 ml/min) is maintained through the pyrolysis tube by fitting a
specially made cap on the end with the ground glass joint. This cap
consists of a 24/40 female joint with 2 septa holes. The carrier gas
line is attached to one end; the thermocouple is placed through the
other. A Type J thermocouple is used with the end of the stick in
contact with the sample bed. The temperature is maintained by an Omega
CN 2000 programmable temperature controller which conducts the pyrolysis
according to the temperature profile in Figure 2-1. A bleach bubbler is
attached to the out gas end in order to deactivate the various Cyano
moieties that are evolved during the pyrolysis procedure.
The various co-pyrolyzed samples discussed in Table 2-2 were
prepared by grinding the appropriately weighted mixtures in a mortar and
pestle and then placing them in the pyrolysis tube as usual. The
samples co-pyroloyzed with NaCl were then stirred in 500 ml of H20
overnight to wash out the salt.
Diffuse Reflectance Infrared Fourier Transform (DRIFT)
spectroscopy was done on a Nicolet 5 DXB FTIR spectrometer fitted with a
Barnes Analytical Diffuse Reflectance unit. Samples were mixed with at
least a hundred fold excess of KBr in order to allow more signal
throughput. The spectra were referenced against a pure KBr background.
Electron Paramagnetic Resonance Spectroscopy was done on a Brukker EPR
operated at room temperature. Elemental Analysis for C, H, and N were
performed by the University of Florida Department of Chemistry
Microanalytical service. Analysis for oxygen content as well as routine
checks on the C, H, and N data were done by Desert Analytics
Microanalytical Laboratory of Tuscon, Arizona.
As mentioned earlier, this sample was purchased from Anderson
Development Company of Adrian, Michigan. The samples were dried in
vacuo at 1000 C for several hours prior to use. The complete elemental
analysis was done by Desert Analytics.
The sample was purchased from Polysciences Company. The pyrolysis
was performed isothermally at 2500 C in either the same apparatus as the
PAN samples or in a standard muffle furnace. Since many of the
applications for the material required the use of pellets of the
composites, samples were pressed in a 1 cm die at 10,000 psi on a Carver
Laboratory Press prior to pyrolysis. Composites made with metallic
species such as Fe(0) or Ti(0) were simply mixed by grinding in a mortar
and pestle. sonically doped composites such as LiC1/PPO were prepared
by co-dissolving the appropriate mixtures of the materials (8% by weight
doping was commonly used) in an azeotrope solvent composed of 7% C2H50H
in CHC13. This azeotrope works quite well because the ethanol dissolves
the inorganic salt and the chloroform dissolves the polymer. When the
solution becomes homogeneous (usually after about 15 minutes of
stirring) the azeotrope solvent is rotovapped off. This method allows
coprecipitation of the dopant and support species and thus affords
The pyrolosis FTIR work was graciously performed by Professor
Thomas Brill at the University of Delaware. The TGA and DSC studies
were graciously conducted by Professor Michael Babich at the Florida
Institute of Technology.
CARBON MOLECULAR SIEVES AS CATALYSTS
This chapter focuses on three different types of substrates and
their reactivity with the CMS catalysts PPAN and AX21. An alkyl
aromatic compound, ethyl benzene, has been studied in the
dehydrogenation reaction to produce styrene. A linear olefin, butene,
has been investigated in dehydrogenation studies to form butadiene.
Lastly, a series of C3 oxygenated compounds -- n- and iso-propanol,
propionaldehyde, and acetone -- have been studied in dehydration and
dehydrogenation reaction schemes. The first two of these catalytic
transformations are large scale industrial processes. In 1987, the
industrial production of styrene was 8,014 million pounds and that of
butadiene was 2,931 million pounds. Estimates for the production in
1989 are even higher; 8,900 and 3,175 million pounds respectively.69
The catalysts typically employed for all of these processes are composed
of inorganic oxides.
The dehydrogenation of ethyl benzene to styrene can be
accomplished by two different catalytic routes. The straightforward
dehydrogenation, shown in equation 3-1below, is used for virtually all
(C6H5)CH2CH3 -> (C6H5)CH=CH2 + H2 AH800K = 124kJ (3-1)
As is shown, this reaction is endothermic and is limited by equilibrium.
In order to achieve acceptable conversions, high temperatures are
required (in excess of 6000 C). Additionally, super-heated steam must
be used as an additive. This provides extra heat needed for the
reaction; reduces the ethyl benzene and hydrogen partial pressures to
maximize yield; and keeps the catalyst clean and active. The catalyst
employed is a promoted iron oxide system.72,73 In a typical reaction
scheme employing this catalyst, conversion is about 50% with over 90%
selectivity to styrene.
In spite of the thermodynamic limitations, there are several
advantages to the straightforward dehydrogenation from an industrial
standpoint. Catalysts lifetimes are very long. Typical industrial
catalysts can last 2-3 years in continuous operation. Catalyst
producers consider a lifetime of least one year as a virtual requirement
for any potential catalyst systems.74 Another advantage to the direct
dehydrogenation is that H2 is obtained as a byproduct. In many
industrial schemes this is used as a fuel to provide some of the heat
required for the reaction.
Despite the long-standing success of the straightforward
dehydrogenation reaction scheme, there has recently been a great deal of
interest in the oxidative dehydrogenation process.75-83 In the presence
of 02 (usually as air), the dehydrogenation is quite exothermic as a
result of the formation of H20 as a byproduct. This reaction is shown
below in equation 3-2.
(C6H5)CH2CH3 + 1/2 02 ---> (C6H5)CH=CH2 + H20 AH0800K= -119 kJ (3-2)
Due to the favorable thermodynamics of this reaction scheme, the process
should be able to operate at lower temperatures and complete conversion
is theoretically possible.
As with the direct dehydrogenation, the systems most often studied
for the oxidative process are based on promoted inorganic oxide
catalysts. One of the major drawbacks of the early systems studied for
this pathway is that selectivities are low. Since the primary use of
styrene is as a monomer for polystyrene production, it is essential that
the selectivity be very high in order that any small amount of side
products can be completely removed. The bismuth molybdenum catalyst,
which is an industry standard for mild oxidations, has a selectivity of
only about 50% in this reaction.78 Recent studies employing other
inorganic oxide systems, however, have shown much improvement with
regards to selectivity. Murakami and co-workers have studied a SnO2
P205 catalyst and found it to have moderate activity--greater than 30%
conversion with greater than 80% selectivity--at 5500 C.78-80 Their
work has particularly focused on the role promoters play in modifying
the acid-base properties of the catalysts. Emig and Hofmann have
reported a Zr02 P205 system which is slightly more active in the same
temperature range -- conversions up to 55% with over 80% selectivity.83
Vrieland has similarly worked with phosphate promoted catalysts and has
recently reported a cerium pyrophosphate system which exhibits 76%
conversion with 90% selectivity at 605 0C.74'81
In addition to focusing on the acid-base properties of the system,
the latter two studies also investigated the role of carbonaceous
overlayers on the surface of the catalyst. It was determined that each
of the various systems studied had an induction period during which a
thin uniform layer of carbon built up on the surface. A subsequent
study suggested that the nature of the layer is a circular pattern of
condensed rings.81 In another recent study, Cadus and co-workers
similarly found carbon overlayers present on a sodium doped alumina
catalyst.82 They provide a mechanism to show how the carbonaceous
species, not the inorganic oxide, bring about the catalytic
Since the actual catalytic surface is primarily carbon and
carbonaceous substances such as charcoals have been known for a long
time to have some dehydrogenation ability,84'85 it is possible that the
bulk of the catalytic reactivity in such systems is a function of the
carbon. The major role of the inorganic oxide constituents may merely
be to modify the acid-base properties of the carbon. This is the
premise for employing CMS catalysts in the oxidative dehydrogenation of
ethyl benzene to styrene.
As mentioned previously, the dehydrogenation of butene to
butadiene is a widely studied industrial process. The typical catalyst
is a mixture of Bi203 MoO3 which is usually referred to as bismuth
molybdate. This system requires a reaction temperature of at least 5000
C to yield a conversion of 40% with 95% selectivity.86 The process is
virtually always done via an oxidative route as shown by the reaction
CH2=CH-CH2-CH3 + 1/2 02 -----> CH2=CH-CH=CH2 + H20 (3-3)
Recently a number of other inorganic oxide based systems have been shown
to be even more active under similar reaction conditions. Lambert and
Germain report that a tin cadmium phosphate catalyst exhibits 88%
conversion with 71% selectivity at 4700 C.87 Herniman and co-workers
have employed a tin antimony mixed oxide catalyst and have observed
The mechanism that is commonly invoked for these mixed oxide
catalysts focuses on lattice oxide ions as the active sites where
adsorption and dehydrogenation of the alkene occur.89 The role of the
gaseous oxygen reactant is to reoxidize the catalyst. This is
accomplished by dissociation on the surface to form oxide ions and then
diffusion through the bulk to the active site.90 It is fairly well
established from kinetic studies that the reaction is zero order in
oxygen at oxygen to butene ratios above 0.5.91
In light of the proposed mechanism for inorganic oxide catalysts
which utilizes the oxide species to a significant extent, one would
initially conclude that CMS based catalysts should not have much
reactivity in butene dehydrogenation schemes. Alternatively, if CMS
systems are active, other mechanistic interactions would seem more
plausible. CMS systems have, however, been shown to be active for
butene conversion and an oxide migration type mechanism has been
proposed. Ademodi and co-workers report that a PPAN based system
oxidatively dehydrogenates butene to butadiene to about 35% conversion
at 3500 C with essentially 100% selectivity.92 The mechanism is not
fully delineated, but it is proposed to involve adsorption,
dissociation, and migration of oxygen species through the sieve
material. Since AX21 is far more porous than PPAN and since it contains
an appreciable amount of oxide functionalities, this should also be an
interesting system to study for the conversion of butene to butadiene.
The final class of substrates to be investigated in this chapter,
the various alcohol and related oxygenated compounds, are not nearly as
attractive from an industrial standpoint as ethyl benzene or butene.
There are far better ways of producing large scale quantities of
propylene, for example, than dehydrating n- or isopropanol. In fact,
this is opposite of the usual direction such catalytic transformations
are pursued--oxygenated products are typically more desirable. The
study of such reaction sequences is nonetheless beneficial. By
examining a variety of substrates, a better understanding of the
reactivity of CMS catalysts can be obtained. Furthermore, the series of
C3 compounds offers the gamut of organic functionalities: straight-chain
alcohol, branched-chain alcohol, aldehyde, and ketone. An analysis of
the reactivity of these catalysts towards the respective compounds, as
well as the nature of the product distributions, should provide a handle
on the mechanistic interactions that are occurring.
The study of the alcohol substrates also offers a further
opportunity to compare the reactivity of CMS catalysts to inorganic
oxide systems. Much work has been done with various A1203 based
catalysts in alcohol dehydrogenation and dehydration reaction
schemes.93,94 The activity is a function of the Lewis acidity of the
catalyst with the products being formed via carbocation intermediates.
Although carbonaceous materials are not usually considered to have much
Lewis acidity, several studies have shown carbon based catalysts to be
active for alcohol substrates with reaction pathways similar to the
inorganic oxide systems. A thorough examination of the reactivity of
these substrates over both types of catalysts should provide a better
understanding of their similarities and differences in catalytic
Ethyl benzene was purchased from Fisher Scientific Company and was
purified by vacuum distillation prior to use. Several later
experiments were performed with ethyl benzene samples that were not
vacuum distilled and the results were the same as in the purified
systems. 1-Butene was purchased from Matheson Gas Products as C.P.
grade with 99.0% minimum purity. Samples were checked by gas
chromatography (GC) to verify that there were no detectable impurities
present. Ethanol, 200 proof, was purchased from Florida Distillers
Incorporated and was dried over 4 A molecular sieves prior to use. n-
Propanol, iso-propanol, and propionaldehyde were all purchased from
Eastman Kodak Company as reagent grade and were dried over 4 A molecular
sieves prior to use. Each sample was checked by GC to verify purity.
Reagent grade acetone was purchased from Fisher Scientific and was dried
in the same manner as the other substrates prior to use. Other
chemicals employed as GC standards for product verification--benzene,
toluene, butadiene, diethyl ether, acetaldehyde--were reagent grade and
used as received.
Preparation of Catalysts
The PPAN samples were prepared as described in the previous
chapter. AX21 samples were dried in vacuo at 100 o C for at least 8
hours prior to use. Unused portions were stored in a desiccator for
future experiments.. The activated charcoal (AC) material was purchased
from Mallinckrodt Incorporated and was dried in vacuo in the same
fashion as AX21. The AC material has a nominal surface area of 800
m2/g. A1203 was chromatographic grade and was purchased from Davisson.
It was calcined at 3500 C prior to use. The nominal surface area of the
A1203 sample is 360 m2/g. The cerium pyrophosphate catalyst was
synthesized from Ce(S04)2 (Alfa) and H3PO4 (Fisher Scientific) according
to the method of Vrieland.74 The industrial mixed oxide catalyst--93%
Fe203, 5% Cr203, 2% K02 on MgO--was prepared according to the method of
Ghublikian.9 All the oxide samples were purchased from Fisher
Scientific and used without further purification.
The reactor is a fixed-bed flow reactor modelled after that
designed and described by Clark.59 For the butene system, the butene
and air reactants are mixed through a two-stage glass bubbler and then
the blended gas stream is fed into the reactor. In all the other
systems, the substrate is delivered as a liquid via a syringe pump. The
pump is also modelled after a design by Clark,59 but additionally is
fitted with an adjustable gear box which enables variable flow rates to
be administered. Temperature is monitored by a thermocouple (type J
connected to an Omega Engineering CN 300 temperature controller ) which
is encased in a glass tube that is in contact with the catalyst bed.
Unless otherwise stated, 0.5g of catalyst is used in each experiment.
Catalyst bed height is approximately 30 mm for all CMS systems. A
schematic diagram of the reactor setup is shown in Figure 3-1.
In the liquid feed systems, a six inch layer of glass beads
(Fisher Scientific) is placed on top of the catalyst to facilitate
complete vaporization of the substrate and to allow thorough mixing with
the carrier gas before contact with the catalyst. Liquid products are
condensed in a trap that is maintained at 00 C. Blank experiments in
which the system was set up with the reactor tube filled with glass
beads were run for all substrates. No activity was observed in any of
these blank systems.
Routine product analysis is done via gas chromatography. Pre-
catalyst and post-catalyst gas samples, as well as post-catalyst liquid
samples, are periodically examined for each experiment. Typical
injection sizes are 0.5 ml for gas samples and 0.15 ul for liquid
samples. For the ethyl benzene experiments, a Varian 3400 GC fitted
with a 3m Dega S column and a Flame Ionization Detector is employed.
The column temperature is ramped from 75 1500 C at 100/min. Data
analysis is performed with a Varian 4290 recorder/integrator. For
butene and the various oxygenated substrates, a Varian 940 GC fitted
----- Carrier gas inlet
---- Tube furnace
-- Reactor tube
---- Septa for gas
---- Dewar filled
Schematic diagram of reactor setup.
with a 6 ft. Hayesep Q column and a Flame Ionization Detector was
operated isothermally at 1300 C. Data analysis is performed on a
Hewlett Packard 3390A recorder/integrator. Further product
identification and verification is done by GCMS using a Varian 3400 GC
interfaced with a Finnegan MAT ITDS 700 Mass Spectrometer. The column
employed for this technique is'a 15m DB1 column operated isothermally at
400 C. Quantification for all GC data is done by standard calibration
methods. Reported conversion percentages are based on the difference
between the moles of substrate entering the reactor and the amount
present in the post reactor stream.
Results and Discussion
Ethyl Benzene Substrate
As mentioned in the background section, one of the major
advantages of the oxidative dehydrogenation process is that it can
operate efficiently at lower temperatures than the direct
dehydrogenation scheme. Nevertheless, most of the recently reported
oxydehydrogenation catalysts have been studied in the 4500-6000 C
temperature range.74-83 Since these systems must first form the active
carbonaceous overlayer--a step most likely requiring a significant
amount of heat--it is quite possible that the CMS catalysts can function
at significantly lower temperatures. Another motivation for operating
at reaction temperatures below 4500 C is that this is the final
pyrolysis temperature of the PPAN samples used in this study. As
discussed in the previous chapter, PPAN can be further pyrolyzed up to
15000 C to go all the way to the formation of carbon fibers. Our
results have shown, however, that the condensed pyridine ring
structure--which is the proposed catalytic species videe infra)--is
maximized at temperatures up to 4500 C. Reaction temperatures above
4500 C will lead to further structural rearrangements and the formation
of volatile surface species which could lower catalyst activity and
complicate product analysis.
The first series of catalytic experiments were performed at
2500 C. Table 3-1 shows the results for various PPAN systems in the
ethyl benzene oxydehydrogenation reaction. The reported product
analyses are calculated from post-catalyst gas samples according to the
method described in the experimental section of this chapter. It is
evident that the PPAN catalysts have slight to moderate activity with
high selectivity to the styrene product. Also shown in Table 3-1 are
the results for an A1203 sample in this reaction scheme. This
experiment is done in order to compare the reactivity of a typical high
surface area (360 m2/g) inorganic oxide under the particular reaction
conditions employed. As is shown in the table, the oxide sample is not
The observation of this type of reactivity at such a low
temperature is quite encouraging. Since the above study is performed at
less than half the temperature commonly employed for this reaction, a
series of studies are done at higher temperatures to see if the activity
can be further increased. The results for the PPAN materials in the
oxydehydrogenation of ethyl benzene at 3500 C are shown in Table 3-2.
It is observed that all systems are far more active at this temperature
Table 3-1 Ethyl Benzene Reactivity Over PPAN Catalysts at 2500 C
R EIRRAC GAS
Substrate delivery--0.2 ml/hr. Carrier flow--5 ml/min.
Reaction time--20 hrs for air runs, 4 hrs for N2 runs.
a) non-styrene products are CO2, benzene, toluene, and benzaldehyde.
b) percentage of all products that are styrene.
SAMPLE CARRIER GAS
Ethyl Benzene Reactivity Over PPAN Catalysts at 3500 C
CONDITIONS: Same as those in Table 3-1.
IZAMD I 1: Lrv~------
than at 2500 C. Since 3500 C is still far below the reaction
temperature of other reported catalyst systems, all further catalytic
studies for this reaction are done at this temperature.
Several observations can be made form the data in Table 3-2. It
is clear that surface area is a key factor to the level of activity.
PPAN3 which was co-pyrolyzed with NH4C1 and thus has a significantly
higher surface area than the undoped PPAN1 sample, is twice as active
for the conversion of ethyl benzene to styrene. This is not at all
surprising since a higher surface area allows more substrate/catalyst
interaction. Such a direct effect between surface area and activity has
been shown for a wide variety of heterogeneous catalyst systems.
It is also evident from the data in Tables 3-1 and 3-2 that this
catalytic process is occurring via an oxidative route. When the system
is operated in the absence of 02 (present in the air carrier flow) the
activity is only slight and it subsides within a couple of hours. The
system can be cycled back and forth, however, between active stage (air
carrier) and inactive stages (N2 carrier). Upon return to an air flow,
reactivity is returned virtually immediately. This series of
experiments was followed through five complete cycles and no loss in
activity or selectivity was observed. The mechanism typically invoked
for dehydrogenation processes over PPAN catalysts involves hydrogen
abstraction from the substrate producing a hydrogenated PPAN moiety.
The catalytic cycle is completed by reaction with 02 to yield H20 and
the original catalyst structure. This reaction scheme was previously
shown in Figure 1-1. The inability of the PPAN to be dehydrogenated is
the reason suggested for the eventual loss of activity in the N2 carrier
Despite the fact that this mechanism has been widely proposed to
account for the catalytic activity of PPAN samples,31'34'41-43 it should
be recalled that PPAN1 and PPAN3 also contain 8.5% by weight oxygen and
oxygenated carbonaceous residues have been proposed as active catalysts
for this reaction. PPAN2 which is partially prepared in air according
to the method of Degannes and Ruthven,43 has an appreciably higher
oxygen content -- 13%. Furthermore the DRIFT results discussed in the
previous chapter show that some of this oxygen content is present in
carbonyl functionalities not unlike the polynapthoquinone moieties
proposed by Cadus82 and Emig and Hoffman.83 In light of this, one might
expect PPAN2 samples to have a higher reactivity than the other PPAN
samples. This, however, is not the case. PPAN2 is only slightly more
active than PPAN1, less active than PPAN3, and has a lower selectivity
than either of the N2-carrier preparations. It is concluded from this
then, that the majority of dehydrogenation activity in the PPAN samples
reported here takes place by the typical mechanism depicted in
Figure 1-1. Any participation by oxygenated moieties is considered to
The direct relationship between higher surface area and higher
activity was the motivation for investigating the commercially available
CMS system prepared from coal pitch residues. AX21 possesses one of the
highest surface areas ever measured -- >2500 m2/g. Additionally, the
novel structure and the oxygen content of this material make it a very
close analogue to the carbonaceous overlayer species present on the
inorganic oxide catalysts discussed previously. The results of these
commercially available CMS materials, as well as the data for samples of
the industrial Fe203/Cr203 catalyst and Vrieland's cerium pyrophosphate
system are shown in Table 3-3.
It is evident that under the experimental conditions of this
study, the two literature-preparation samples are not very active. This
is not unexpected since neither of these systems is reported to have
much activity at 3500 C. AX21, on the other hand, is remarkably active.
Conversion of 80% with greater than 90% selectivity is observed in a
single pass. Like the PPAN systems, this is an oxidative process which
can be cycled. When run in N2 carrier flow, the activity drops
immediately. When air is returned, the formation of styrene is readily
observed. This system is also active over a long period of time. An
extended run was followed for 120 hours in continuous operation and no
appreciable loss in conversion or selectivity is seen. The sample can
even be removed from the system for several days and activity is re-
established when the reactor is set up again. Pre-catalyst and post-
catalyst weight is further monitored in order to determine if the sample
itself is being oxidized under the reaction conditions. No appreciable
catalyst weight loss is observed in any of the AX21 runs described here
for the oxydehydrogenation of ethyl benzene.
Clearly, a significant amount of this increased reactivity is due
to the extraordinary surface area of the AX21 material. This alone,
however, can not explain the overwhelming activity. The activated
charcoal (AC) system has a surface area of 800 m2/g, and yet its
activity is only slightly greater than PPAN3, the "high surface area"
Table 3-3 Ethyl Benzene Reactivity Over AX21 and Other Catalysts at
SAMPLE %CONVERSION %SELECTIVITY
AX21 80.0 90.1
A.C.a 26.0 84.2
CeP207b 6.7 92.5
Fe/Cr/K/Mgc 0.9 45.1
AX21 (N2 flow) 5.2 40.4
CONDITIONS: Same as those in Table 3-1.
a) Activated Charcoal
b) Prepared according to reference 74.
c) Prepared according to reference 95.
PPAN material (50 m2/g). The microscopic structure of AX21, which is
composed of carbonaceous lamellae, gives this material an exceptional
adsorption capability. Toxicological studies have shown that these
materials adsorb considerably more substrate than other highly adsorbent
carbonaceous materials.48-50 This affinity for substrate, together with
a surface chemical structure which is especially suited for this type of
catalytic transformation (due to the oxygen content) surely contributes
to the outstanding reactivity of AX21 in this reaction. The specific
mechanism for the AX21 catalyst is most likely analogous to that
proposed by both the Cadus82 and Emig83 groups. Shown in Figure 3-2,
this reaction scheme invokes a concerted hydrogen abstraction from the
substrate by the oxygen functionalities of the carbonaceous overlayer
formed from coke deposition on an inorganic oxide surface.
Although there is not near as much evidence on the role of
carbonaceous materials in the dehydrogenation of butene as there is for
ethyl benzene, there have been several reports of CMS type materials
being used in this reaction scheme. Table 3-4 shows the results for the
dehydrogenation of butene to butadiene over a variety of catalyst
materials. It is evident that only the AX21 sample has any appreciable
activity under the reaction conditions employed. The A1203 sample is
essentially unreactive and as before, this is not unusual at these low
reaction temperatures (2500 C). It is surprising though, that the PPAN
systems are not more active in this reaction scheme since Ademodi and
I 0 1 0
H20 -- 1/2 02
Proposed mechanism for AX21 dehydrogenation activity.
Table 3-4 Butene Reactivity Over CMS Catalysts
Reaction temperature--2500 C.
Carrier feed--3 ml/min.
Substrate feed--2 ml/min.
Percentage of products which is butadiene. Other products are C1 -
co-workers reported a PPAN system that has 20% conversion at 2500 C92
The Ademodi catalyst is prepared in a slightly different manner and the
authors do not present any characterization studies. Another difficulty
is that some of the reactor data, such as catalyst bed height, is not
given. As discussed in Chapter 2, there are some problems with
comparing PPAN samples from one author to the next.
The results for AX21 show that this material has moderate activity
with excellent selectivity to butadiene. The conversion of 11% appears
low compared to the reactivity of this catalyst towards ethyl benzene,
however, other studies have shown carbonaceous materials to have less
reactivity in this system than in ethyl benzene systems. Emig and
Hoffman report considerably lower reactivity for their carbonized
zirconium phosphate system in butadiene conversion as opposed to ethyl
In order to extend the study of the catalytic reactivity of CMS
materials, a series of oxygenated substrates are investigated. Initial
work is done with ethanol. The dehydrogenation of ethanol to
acetaldehyde is well characterized on inorganic oxide catalysts.96'97
The results for the reactivity of this substrate over CMS catalysts at
2300 C is shown in Table 3-5. Again the results for an alumina run are
also included to directly compare the reactivity of CMS materials to
commonly employed inorganic oxides. At 2300 C, the A1203 sample is
found to have very little activity. The PPAN catalyst (PPAN3, the high
surface area sample) is considerably more active. Overall conversion is
Ethanol Reactivity Over CMS Catalysts
SAMPLE CARRIER ACETALDEHYDEE
%ETHYL %DIETHYk %OVERALL
ACETATE ETHER CONVERSION
Reaction temperature--2300 C.
ml/hr.Carrier flow--5 ml/min.
air runs, 4 hrs for N2 run.
Reaction time--20 hrs for
Percentage of stated product among all products.
20% with the major products being acetaldehyde and ethyl acetate. The
former product is the result of the direct dehydrogenation of ethanol.
The latter product must occur via a secondary condensation reaction in
which an acetaldehyde fragment reacts with an unconverted ethanol
molecule. As with the other substrates studied in this chapter, the key
mechanistic step involved here is hydrogen abstraction. The extensive
work in this area by Iwasawa proposes that the alcohol adsorbs
dissociatively to form a bound alkoxide.96 This is followed by hydrogen
abstraction to produce the dehydrogenated product. PPAN has been shown
previously to be efficient at hydrogen abstraction with alcohol
substrates.31,41,97 The nature of the product profile, acetaldehyde
being the major product, shows that the catalyst preferentially
abstracts a hydrogen from the methylene carbon.
As might be expected from the ethyl benzene and butene systems,
AX21 is far more active than the PPAN system. Overall conversion is 70%
with a product profile very similar to the PPAN catalyst. The major
products are again acetaldehyde and ethyl acetate. The similarity in
the selectivities between PPAN and AX21 towards ethanol conversion
suggest that the two materials operate through similar mechanisms. The
significantly greater activity of AX21 over PPAN, however, exemplifies
how much stronger a hydrogen abstraction catalyst is the former CMS
The ethanol results are enlightening in that some reactivity
information is obtained, but since there is really only one possible
hydrogen abstraction site (it would be highly unlikely for any
abstraction to occur at the methyl hydrogens) not much mechanistic
information can be learned. For this reason a series of C3 substrates
are examined. The straight-chain alcohol, n-propanol, the branched-
chain alcohol, isopropanol, the aldehyde, propionaldehyde, and the
ketone, acetone, are each studied for their reactivity with the AX21
The results for the C3 substrates are shown in Table 3-6. The
reaction temperature (2300 C), catalyst weight (0.5g), carrier flow
(5ml/min), and substrate flow (0.2 ml/min) were all the same as for the
ethanol experiment. It is evident that the alcohols were quite
susceptible to both dehydration and dehydrogenation. The aldehyde
substrate was equally reactive, however, the products are significantly
different--they are all C2 species. Acetone is virtually unreactive as
a substrate. Analogous to the ethanol system, the reactivity for the C3
substrates is dependent on the nature of the carrier flow. When N2 is
used in place of air, product formation ceases.
For n- and iso-propanol, the major product is propylene. Proposed
mechanistic pathways to this and the other products observed for the
former substrate are shown in Figure 3-3. The latter substrate reacts
via similar routes. As is shown, the initial coordination of the
propanol occurs via the donor oxygen to a Lewis acid site on the
catalyst surface. This mode of coordination is often invoked in alcohol
oxidation processes.96'98 The formation of propylene is the result of
hydrogen abstraction from the secondary carbon (labeled C2 in
Figure 3-3) by a surface oxygen species. This step is drawn in Figure
3-3 as a hydride abstraction which produces a carbocation. The
formation of the alkene is shown to occur in a subsequent step. It is
Reactivity of C3 Oxygenated Substrates Over AX21 Catalyst
L LAREVO CONV ERS IONb
Reaction temperature--2300 C.
Substrate flow--0.2 ml/hr.
Carrier flow--5 ml/min.
Percentage of stated product among all products.
Other products include small amounts of ethers and esters.
C C CH
/ H H
0 0 0O
\\\\^\\\\^^ ~ ~ ~ \ Iu I, ''i u Ji~ im I
\ 02 Abstraction
C C C CH3
H ?'H 3
0C+ C- CH3
HO 0 HO
HO 0 / OH
Proposed mechanistic routes for conversion of
quite possible that a concerted process which incorporates both of these
stages is occurring. This is analogous to the proposed mechanistic
pathway in the ethyl benzene system. The formation of propionaldehyde
must occur through abstraction from the functionalized carbon (labeled
C1 in Figure 3-3). The fact that a secondary carbocation is more stable
than a primary one most likely accounts for the predominance of the
alkene product. The observation of a small amount of di-n-propylether
shows that coupling reactions can also occur. It should additionally be
noted that there is a small amount (<2%) of acetaldehyde and ethylene
produced in the n-propanol system. These products are formed via
secondary reactions of the propionaldehyde species videe infra). This
shows that the catalyst is able to readsorb and cause further reaction
of the primary products.
As is shown in Table 3-6, the propionaldehyde substrate yields all
C2 products. This seemingly unusual result can best be explained by a
hydrogen atom abstraction mechanism. Abstraction of the aldehydic
hydrogen atom gives rise to and RCO' species. This then decarbonylates
to produce the C2 fragment, R', which on further reaction accounts for
the various products observed. Again this is most likely a concerted
process since attempts to detect a radical intermediate have proved
These results demonstrate that the AX21 material can function in
either hydride or hydrogen atom abstraction reactions depending on the
nature of the substrate. Hydrogen atom abstraction occurs rather easily
on an aldehydic species such as propionaldehyde; whereas, hydride
abstraction is more facile on alcohol substrates. In the case of
acetone, neither hydride or hydrogen atom abstraction should occur
readily and as expected, this substrate is found to be unreactive.
CARBON MOLECULAR SIEVES AS CATALYST SUPPORTS
This chapter focuses on the study of CMS materials as supports for
three different types of heterogeneous catalytic systems. In the first,
PPAN and AX21 have been doped with ruthenium species and have been
investigated in the Fischer-Tropsch reaction. A great deal of work has
been done on this system employing inorganic oxide supports. Virtually
all of these catalysts yield broad product distributions. The goal of
this study is to devise a system which will be active towards the
conversion of synthesis gas (CO and H2) with high selectivity to small
molecular weight products. The second catalyst system studied employs
Mo03 doped AX21 for the oxidation of methanol. As discussed in the
previous chapter, AX21 itself is active towards alcohol substrates.
Methanol, however, is one of the most difficult alcohols to
catalytically transform. The objective of this study is to combine a
catalytically active metal species with a similarly active support to
produce a system which can display a synergism between its constituents.
Lastly, molybdenum and tungsten doped AX21 systems have been
investigated in the deep oxidation of chlorinated hydrocarbons. There
is a growing interest in developing catalytic systems which can fully
oxidize various organic pollutants, e.g. CH2C12, to compounds that are
easily disposed like CO2 and HC1. The goal of this study is to combine
the exceptional adsorption capabilities of the support with the
oxidative catalytic ability of the metal dopant. In each of these
applications, the specific structure and properties of the CMS materials
offer unique advantages towards accomplishing the prescribed goals.
Fischer and Tropsch first reported the catalytic conversion of
synthesis gas to higher hydrocarbon products in 1926.99 During the more
than 60 years since that time, there has been a great deal of work
conducted in the field. Traditionally this reaction has been performed
under rather severe conditions--pressures of 100-200 atmospheres and
temperatures in excess of 4000 C. Throughout the study of this process,
though, there has been a wide range of conditions employed. Table 4-1
lists some of the different systems that have been investigated.
Included in Table 4-1 are the conditions for the SASOL plant in South
Africa which is the only presently operating industrial facility that
uses the Fischer-Tropsch reaction for the production of hydrocarbons.
It is evident that many of the transition metals show some
activity for CO hydrogenation. In considering a industrially viable
system, however, overall activity is not the most significant parameter.
Selectivity to a certain desirable range of products is of paramount
importance. Virtually all Fischer-Tropsch systems eventually give rise
to a Schulz-Flory product distribution.100,101 This model, which was
originally derived to explain polymerization sequences, can be described
by the equations below:
Wn = n (1-a)2 an-1 (4-1)
log Wn/n = n loga + log [(1-a)2/a] (4-2)
Table 4-1 Commonly Studied Fischer Tropsch Systems
Typical Metal Dopants
Ru, Rh, Os, Fe, Co
Si02, A1203, MgO, Ti02
Typical Reaction Conditions
Pressure: 100 200 atm
Temperature: 300 4000 C
1 2000 atm
re: 150 4500 C
220 3400 C
The variables in these equations are defined as follows: n is the carbon
number of a particular polymer unit or, in the case of Fischer-Tropsch
systems, it is the number of carbons in a particular product; Wn is the
weight fraction of that product among all the products; and a is the
probability of chain growth. This parameter is a function of the degree
of polymerization and rates of propagation and termination.
Equation 4-2 shows that a linear function should be obtained from
a plot of log Wn/n vs n. Such a plot is called an Anderson-Schulz-Flory
(ASF) graph.102-104 The slope of the line is related to a which in
Fischer-Tropsch terminology describes the chain growth of the reaction
process. As the a value increases, so does the chain length of the
hydrocarbons formed. This produces a broader product distribution.
Hydrocarbons up to a chain length of C30 are not unusual in many active
Fischer-Tropsch systems. Since Equation 4-2 is derived from a polymer
model, there are certain deviations from linearity when it is applied to
the Fischer-Tropsch process. CI values always fall above the line and
typically are not considered part of the data set. The model is
designed for the linking of methylene fragments such as would occur in a
growing polymer chain. Since CH4 is formed by direct hydrogenation of a
surface carbide fragment, its production is not related to the polymer
model. Many experimental systems also have C2 and C3 values which
deviate somewhat from linearity. Above C4 though, most systems fit ASF
plots very well.
In the aftermath of the oil crisis in the mid 1970's, there was an
explosion of interest in alternatives to petroleum based routes for
hydrocarbon production. Much effort was focused on Fischer-Tropsch
systems for the formation of gasoline range products.71,105,106 Many
systems, most of which employed Group VIII metals supported on high
surface area inorganic oxides, were reported to be highly active.
Despite the research enthusiasm, however, the production of gasoline
range products via CO hydrogenation never became significantly cost
competitive with petroleum pathways. It now seems apparent that, at
least for the foreseeable future, the only realistic market for the
Fischer-Tropsch process is in the selective production of small
molecular weight hydrocarbons and oxygenated products which can be used
for the production of specialty chemicals. It would be particularly
desirable to make alkenes such as ethylene and propylene by this route.
The problem with designing a system to selectively produce the
products mentioned above has already been pointed out. Most CO
hydrogenation systems exhibit a broad product distribution with limited
selectivity towards the low molecular weight compounds. As mentioned
previously, there is a great deal of similarity in the reactivity of the
many metal species studied. The mechanism most often invoked for low
selectivity is related much more to the oxide support than it is to the
metal dopant.107-110 Initially, inorganic oxides such as silica and
alumina provide excellent dispersion of the metal species. Since there
is no great chemical attraction between the oxide or hydroxyl nature of
the support and the metal dopant, however, the metals tend to migrate on
the surface of the support during the course of the reaction. This
migration produces a significant growth in metal crystallite size. Such
a phenomenon causes only a slight decrease in activity, but it has a
detrimental effect on the selectivity. The sintering of metal species
leads to very broad product distributions.108-110 To devise a Fischer-
Tropsch system for the selective production of light hydrocarbon
products then, it seems prudent to focus on the choice of a proper
support as opposed to the design of new metal species. Prospective
supports should have the ability to strongly interact with small metal
clusters and through this interaction, prevent the formation of large
This criteria was the motivating factor for employing PPAN and
AX21 as supports in heterogeneous Fischer-Tropsch systems. The
condensed pyridine ring functionality of PPAN should provide a much
stronger interaction in an acid-base sense than the oxide and hydroxyl
species of inorganic oxides. AX21 should likewise be able to inhibit
cluster formation, but for a completely different reason. The pores and
cavities associated with the structure of this material should provide
isolation of small metal clusters in much the same way as alumino-
silicate zeolite materials have been employed.111-114 Both of these
support systems have the additional advantage of being catalysts
themselves. As was suggested earlier, olefin products are more desired
than paraffins. The dehydrogenation ability of these CMS materials,
such as that discussed in the previous chapter enables the possibility
of synergistic interactions between the metal dopant and the
In addition to the design of Fischer-Tropsch systems with improved
selectivity, mechanistic studies of the basic Fischer-Tropsch reaction
and systems active for CO2 reduction have also been investigated with
CMS supports. In the former work, acetaldehyde is used in a co-feed
with syn gas. By following the fate of this additive, some very
interesting information is obtained regarding the fundamental reaction
steps of the Fischer-Tropsch process. Lastly, these systems are shown
to have appreciable activity towards CO2 reduction. Studies conducted
with CO/CO2 co-feeds further demonstrate how product selectivity can be
The oxidation of methanol is one of the most widely studied
industrial catalytic processes. It is well documented that there are
two main reaction pathways for this substrate.115 The predominance of
one over the other depends mostly on reaction temperature. In the so-
called low-temperature scheme, the major products are methyl format
(MF) dimethoxymethane (DMM) and dimethyl ether (DME). In the high-
temperature systems, formaldehyde can be produced with high selectivity.
This latter process is performed on a major industrial scale.
Formaldehyde is projected to be the fifth largest produced organic
chemical in 1989.69 The relevant reactions and conditions of these two
processes are shown in Table 4-2.
Although the primary focus in methanol oxidation is on the high
temperature production of formaldehyde, there is a growing interest in
low temperature systems which can selectively produce methyl format.
This product is formed through a dimerization of formaldehyde units in
the so-called Tischenko reaction.116-118 Methyl format has been shown
in a number or recent studies to be a versatile intermediate for the
synthesis of such high volume products as acetic acid,119-121 dimethyl
formamide,122 and ethylene gylcol.123
Typical Methanol Oxidation Reaction Routes
1. Low Temperature Process
3 CH3OH +
------> CH3OC(O)H +
Reaction temperature--200-2400 C
Reported catalysts--Mo03, Cu2Cr2O4
86% selectivity (MF)
2. High Temperature Processes
CONDITIONS: Reaction temperature--600-8000 C
Reported catalysts--FeMoO4 or Ag
In light of the reactivity of AX21 toward alcohol substrates which
was discussed in the previous chapter, this catalyst is employed for
methanol oxidation studies. Since a great deal of work had already been
conducted in the low temperature regime, initial studies are done at
2300 C. The goal is to design a system which will exhibit higher
reactivity with appreciable selectivity to the desired product--under
these conditions, methyl format. Higher reaction temperatures are also
studied in the attempt to produce formaldehyde. Both the undoped AX21
and a variety of Mo/AX21 systems are found to be active for the
conversion of methanol. The doped systems are found to be especially
interesting due to an apparent synergism between the metal and the
support. This phenomenon is unique in that it is not observed for
inorganic oxide supported systems. A series of surface science
techniques have been conducted on the Mo doped AX21 catalysts as well as
similarly prepared Si02 supported systems. Scanning Electron Microscopy
(SEM), X-ray Diffraction (XRD), and X-ray Photoelectron Spectroscopy
(XPS) have revealed some insight into the nature of the doped metal
species and how it changes during the course of the reaction.
The last section of this chapter focuses on the deep oxidation of
chlorinated hydrocarbons. The industrial production of these materials
for use as solvents, pesticides and a host of other applications has
been rising continuously over the past ten years. Due to the widespread
use of these materials, there is a growing environmental pollution
problem associated with their improper and inefficient disposal.124
These compounds prove to be particularly difficult to dispose of because
they do not burn efficiently in typical waste incineration plants. This
problem is further complicated by the fact that the by-products formed
from incomplete incineration are often more hazardous than the original
starting material. In light of this, there is a great deal of interest
in catalytic systems which can fully degrade chlorinated hydrocarbons to
much more easily handled materials such as CO2 and HC1. The systems
most often studied for such reactivity involve transitional metal oxides
doped with a variety of constituents.125 Recently an electrolytic
oxidation system was reported to be efficient for a variety of
chlorinated hydrocarbon species including aromatic systems.126
Several molybdenum and tungsten doped AX21 systems are studied for
their reactivity towards CH2C12. This substrate is chosen for two
reasons. First, its conversion should be relatively easy to follow.
Secondly, it is a relevant system to study, since CH2C12 is by far the
largest industrial chlorinated hydrocarbon pollutant.124 The use of the
AX21 support should offer several advantages over unsupported systems or
those employing inorganic oxide supports. Much of the work discussed so
far has shown that AX21 offers a significantly higher and more stable
degree of dispersion than inorganic supports. This increased surface
area of the active metal species should improve the reactivity of this
system in much the same way as it does for other catalytic reaction
schemes. Additionally, the extraordinary adsorption capability of the
AX21 support can greatly enhance the reactivity of the system. This
should especially be true in "real-world" operations where the organic
pollutant is only present in small concentrations. A lot of work is
being done in employing carbonaceous materials for ground water
purification.47 The work described in this section thus combines the
two current methods for toxic waste clean up. The CMS support is
employed to more efficiently adsorb the specific pollutant from the
waste stream and the metal dopants are present for the deep catalytic
oxidation of these materials.
Synthesis Gas Conversion
The reactant gases were delivered directly from their respective
cylinders and were not further purified. The carbon monoxide (Matheson
Gas Products) used was Coleman Purity grade with a minimum purity of
99.5%. The carbon dioxide (Matheson) sample was Coleman Instrumental
grade having a minimum purity of 99.99%. The hydrogen (Liquid Air
Corp.) supply was prepurified grade with a minimum of 99.9%. The
ruthenium metal dopings were prepared from both RuC13 (Alfa) and
Ru3(CO)12 (Strem Chemical Company). The acetaldehyde (Fisher
Scientific) reactant was dried over 4A molecular sieves prior to use and
was stored in the chemical refrigerator. Identification of products via
gas chromatography was aided by the use of Scotty Analyzed Gases (Scott
Specialty Gases Company). Three different samples were employed: Can
Mix 1 containing C1 C6 linear paraffins; Can Mix 3 containing C2 C6
normal olefins plus acetylene; and Can Mix 32 containing CI C4 linear
paraffins plus isobutane. Isotopic labelling studies were conducted
using a 25% 13C labelled CO sample (Isotec Incorporated, Dayton Ohio).
The methanol reactant (Eastman Kodak Company) was spectro grade
and was dried over 4A molecular sieves prior to use. The absence of
impurities was verified by gas chromatography. The air supply (Liquid
Air) was U.S.P. breathing quality. The various metal dopants employed
were reagent grade and were used without further purification. Mo03 and
Na2MoO4 were purchased from Aldrich Chemical Company. The ammonium
molybdate sample was purchased from Mallinkcrodt under the name of Acid
Molybdic. This commonly used reagent is nominally (NH4)2MoO4 but is
essentially 85% Mo03
with added NH40H. The standards for product identification--Methyl
format (Fisher), dimenthoxymethane (Chem Service Company, West Chester,
Pennsylvania), and dimethyl ether (Matheson)--were reagent grade and
used without further purification.
Deep oxidation of chlorinated hydrocarbons
Methylene chloride (Mallinkcrodt) was reagent grade and was used
without further purification. The absence of impurities was verified by
gas chromatography The various metal dopants were reagent grade. The
molydenum species and the air supply were the same as those in the
methanol oxidation studies. All the tungsten samples -- WO3, Na2WO4,
(NH4)2W04 were purchased from Aldrich. The molecular formula for the
last species is not completely accurate. The supplier provided
elemental analysis data which reported the W content to be 66.9% by
weight. The simplified formula given above corresponds to 64.79%
tungsten. The value give by the supplier was used when calculating
molar quantities for catalyst preparation.
Preparation of Catalysts
The preparation of the PPAN and AX21 support materials was
described in Chapter 2. The AX21 sample was dried for eight hours in
vacuo at 1000 C prior to use. This is done in order to obtain an
accurate weight of the sample. The ruthenium doped catalysts for
synthesis gas conversion were prepared by combining 0.05g of Ru3(CO)12
or RuC13 with 2.45g of PPAN or AX21. For either dopant, this loading
corresponds to approximately 1% by weight Ru metal--0.98% in the case of
RuC13 and 0.96% for Ru3(CO)12. For the Ru3(CO)12 doped samples, the
mixed powders were refluxed in 60m1 of hexane (a mixture of isomers) for
a minimum of six hours. The term reflux is not totally accurate since
only the Ru3(CO)12 dissolves. This heating and mixing process, however,
serves to better disperse the metal cluster onto the support. For the
RuC13 doped samples, a similar reflux step is employed using H20 as the
solvent. After the heating and mixing process is completed, the
respective solvents are removed by rotary evaporation. The catalysts
are then dried in vacuo at 600 C (1000 C for the H20 systems) for eight
hours. A Ru3(CO)12 doped A1203 sample used in comparative catalytic
studies was prepared analogous to that of the PPAN and AX21 supported
In addition to the solvent deposition route described above, doped
PPAN samples were also prepared by mixing the Ru species and PAN
together before pyrolysis and firing the combined mixture. This
technique did not prove to be very successful though, since none of the
catalysts prepared in this manner exhibited any reactivity. It is
believed that this method is unproductive because the metal dopant gets
buried in the interior of the support matrix and is thus not readily
accessible to reactant gases.
The presence of the Ru3(CO)12 cluster on the PPAN and A1203
supports can be verified via DRIFT Spectroscopy. Figure 4-1 shows the
IR spectra of PPAN before and after being doped with Ru3(CO)12. The
metal carbonyl stretching frequencies can be observed in the 2000-2200
cm-1 region. Figures 4-2 and 4-3 display the metal carbonyl stretching
frequency region for unsupported Ru3(CO)12 and Ru3(CO)12/PPAN. It is
evident that the peaks for the supported system are shifted to lower
wavenumbers indicating a bonding interaction with the support. There
has been a great deal of work in relating the position of the peaks to
the type of the metal/support interaction and to the nature of the metal
species.127-130 The Ru/A1203 system, for example, has been well
characterized by IR assignments. The IR spectra for the Ru/PPAN system,
however, are not suitable for this type of detailed analysis. The peaks
are very broad and due to the high level of noise inherently associated
with PPAN IR spectra, inferences drawn from exact peak locations can not
The various molydenum and tungsten doped samples were prepared by
the same solvent deposition method. In all cases, deionized H20 was
used. For the Mo03/AX21 sample, 0.5 g of the metal oxide was mixed with
2.5g of dry support. This loading corresponds to 3.47 x 10-3 moles of
Mo and 11% by weight metal content. This level of loading was chosen
because most literature work on support Mo03 systems is in the 8-12%
range.131-133 In order to maintain a degree of consistency between the
I I I I
4000.0 3400.0 2800.0 2200.0 1600.0 1000.0 400
DRIFT spectra of PPAN and Ru3(CO)12/PPAN.
I I -I ---I
2200.0 2100.0 2000.0 1900.0 1800
X- 1933. 8 Y 84. 541
X- 1949 6 Y= 72.091
X= 1964. 4 Y= 69. 826
X= 1980. 9 Y= 55 733
X= 1987 7 Y= 50. 699
X= 2001.0 Y= 44 523
X- 2019 6 Y= 43 876
X= 2028.9 Y= 42 688
X= 2041. I Y= 45.250
X= 2062. Y= 45. 146
X= 2066. 2 Y= 45. 591
X= 2119.6 Y= 84 696
DRIFT spectrum of unsupported Ru3(CO)12.
I r --1-- ---- ------
2200.0 2100.0 2000.0 1900.0 1800
X= 1960 2 Y= 13 192
X= 1985. Y= 12 927
X= 1994 8 Y= 13 093
X= 2016. 3 Y= 13 207
X= 2022.5 Y= 13 041
X= 2044.5 Y= 13 020
X- 2052. 4 Y= 12 901
DRIFT spectrum of Ru3(CO)12/PPAN.
various metal dopant precursors, this molar value was used for all the
other Mo and W species. For the Na2MoO4 sample, for example,
3.47 x 10-3 moles of Mo corresponds to 0.71g, so this amount is combined
with 2.45g of AX21.
The reactor for all of the studies conducted in this chapter is a
fixed bed flow reactor of the same type as that employed in the
experiments of the previous chapter. Unless otherwise stated, 0.5g of
catalyst is used in all experiments. For the synthesis gas conversion
experiments, gas delivery was metered via a three stage silicon oil
bubbler fitted with rotoflow valves. The design of this device has been
described elsewhere.134,135 The top two stages were always used for H2
and CO flow respectively. The bottom could be used for N2 purging, CO2
reactant flow, or for HC1 reactant flow. When the latter species was
employed, concentrated sulfuric acid was used in place of the silicon
oil. The total flow rate of gas reactants was normally maintained at
4ml/min. In most syn gas experiments the ratio of H2/CO was 3:1.
Several studies were done with higher CO levels up to a H/2CO ratio of
For the acetaldehyde co-feed experiments, an additional bubbler
was fitted directly to the top of the reactor tube just outside the
oven. Connection to the reactor tube was afforded via a ground glass
joint. The bubbler was filled with 3m1 of acetaldehyde. In order to
slow the rate of substrate delivery, the bubbler is placed in a Dewar
and the temperature is maintained at 00 C. The quantity of the aldehyde
going into the reactor can be monitored chromatographically through a
Reaction temperature for all syn gas systems is 2200 C. This
temperature is monitored by a thermocouple encased in a glass tube in
contact with the catalyst bed. The design of the oven and the other
specifications of temperature control are the same as those in the
experiments described in the previous chapter.
In the methanol oxidation studies, the substrate is delivered via
the syringe pump described previously. The flow rate is maintained at
0.2ml/hr. Carrier gas flow, either N2 or air, is monitored by a Cole
Parmer variable flow controller. The flow rate is set at 5ml/min. In a
typical experimental cycle, the initial heating stage is done in N2 flow
with the syringe pump turned off. After reaction temperature is
reached, 2300 C unless otherwise stated, post gas samples are taken to
verify that no thermal decomposition products are present. At this
point carrier flow is switched to air. Post gas samples analyzed by gas
chromatography show that it takes about two hours for the N2 flow to be
completely purged. After the air carrier flow is determined to have the
proper N2/02 ratio, the syringe pump is turned on. The post reactor
effluent is collected in a trap which is maintained at 00 C.
The air injections prior to substrate delivery also serve to
detect any CO or CO2 being produced by oxidation of the CMS material
itself. No CO is observed in any of the systems. Trace levels of CO2
are observed at 2300 C. Usually, the CO2 peak is not even large enough
to be integrated and thus is too small to be accurately quantified. At
high temperature, however, this formation of CO2 from the oxidation of
the CMS material is greatly increased. These observations are discussed
in depth later in this chapter.
The experimental design for the deep oxidation studies is very
similar to the other systems previously discussed. The substrate is
delivered via the pre-reactor bubbler attached directly to the top of
the reactor tube. The bubbler is filled with 5 ml of CH2C12. As in the
acetaldehyde experiment, the delivery is slowed by cooling the bubbler
in an ice bath. The carrier flow rate is also reduced so that only a
small amount of CH2C12 is being delivered. The flow rate employed is
less than 1 ml/min. The post reactor effluent is condensed in a trap
which contains 10 ml of deionized H20. The purpose of this quenching in
H20 is to dissolve the HC1 that is generated from the deep oxidation.
The amount of acid formed can be quantified by titration with NaOH.
Periodic gas sampling of the post effluent also establishes the activity
of the system since CO2 formation can be observed.
For all three of the heterogeneous catalytic systems discussed in
this chapter, bed height and catalyst weight are measured before and
after each experiment. Very little loss in either bed height or weight
is observed in systems run at temperature below 2500 C. Above these
temperatures, however, appreciable degradation of the CMS material
occurs. This will be discussed further later in this chapter.
Routine product analysis for all the systems discussed is done by
Gas Chromatography. Both Flame Ionization Detection, suitable for
hydrocarbon analysis, and Thermal Conductivity Detection, efficient for
N2, 02, CO, CO2 and HC1 analysis, are employed. The former was
conducted on a Varian 940 GC fitted with a Hayesep Q column. The latter
technique was conducted on a Hewlett Packard 5700A GC employing a
Hayesep DB column. Quantification of data was done by a Hewlett Packard
3390A integrator/recorder. Samples are taken in both the gas phase at
the exit of the reactor oven and from the liquid phase after the
effluent has been condensed in the cold trap. Further product
identification was offered through the use of GCIR and GCMS techniques.
The latter method was already described in the preceding chapter. The
former was accomplished with a Hewlett Packard 5890A GC using a DB 130
column in conjunction with a Nicolet 5 DXB FTIR.
Some of the products formed in the systems discussed here are not
amenable to GC detection and quantification. In light of this, a couple
of titration methods have been employed. To quantify formaldehyde
produced in the high temperature methanol oxidation studies, the sulfite
titration method of Walker is practiced.136 The reaction involved for
the titration is shown below:
CH20 + Na2SO3 + H20 -----> NaOH + CH2(NaSO3)OH. (4-3)
The method is done by adding a few drops of phenolphthalein
indicator (0.5g phenolphethalein in 50m1 ethanol) to a 1.0m solution of
Na2SO3 (12.6g Na2SO3 in 100ml D.I. H20). The solution turns magenta
because SO32- (aq) is a base. The color is returned to clear with the
addition of several drops of a standard 1.0 N HC1 titrant solution. The
next step is to add a weighed amount of the reaction solution to be
titrated. The flask is shaken to allow complete mixing. The solution
turns magenta again to indicate that base is present. The quantity of
NaOH(aq) generated is determined by titration with the HC1 solution.
The % formaldehyde present in the sample can be readily calculated from
the equation below:
% formaldehyde = ml Acid titer x Normality of acid x 3.003 (4-4)
1 x Weight of sample
The constant, 3.003, comes from the fact that one ml of 1.ON acid is
equivalent of 0.03003g of CH2O.
As mentioned above, the quantification of HC1 produced in the deep
oxidation studies is also determined titrametrically. A standard 1.0 M
NaOH solution is used with a phenolphthalein indicator. The number of
moles of HC1 produced is equivalent to twice the number of moles of
CH2C12 oxidized according to the balance equation shown below:
CH2C12 + 02 ---> 2HCL + CO2 (4-5)
Characterization of the catalytic materials is conducted by a variety of
spectroscopic techniques. DRIFT Spectroscopy is performed as was
previously described in Chapter 2. Scanning Electron Microscopy is
performed with a Jeol JSM 35C electron microscope using an acceleration
voltage of 25 kV. Samples are initially examined at 300X in order to
insure the regions to be studies at higher magnification are
representative of the whole. Micrographs are taken at 6000X and
40,00X. The author would like to acknowledge Richard Krockett of the
University of Florida's Major Analytical Instrumentation Center for his
assistance with the SEM studies. X-ray Diffraction Spectroscopy (XRD)
was run on a General Electric 3000 instrument using Cu a radiation. The
author acknowledges the research group of Dr. T.E. Mallouk at the
University of Texas at Austin for conducting the XRD studies. X-ray
Photoelectron Spectroscopy (XPS) was run on a Kratos 9000 instrument.
Samples were mounted on aluminum tape and were evacuated for eight hours
prior to scanning. The author acknowledges Dr. Vaneica Young and Mr.
Michael Clay of the University of Florida for their assistance in
performing these XPS studies. For each of the surface techniques
mentioned above, SEM, XRD, and XPS, the exact same samples were used.
Results and Discussion
Synthesis Gas Conversion
As mentioned in the introduction to this chapter, most of the
transition metals exhibit some activity towards CO hydrogenation.
Ruthenium was chosen as the metal constituent in this study since it has
been shown to be very active for both CO101'127'129'137-146 and CO2
reductions.147-150 Initially, work was conducted with RuC13 as the
metal precursor. For most synthesis gas conversion systems, especially
those operating at ambient pressures, the active metal species is zero
valent. Thus, this Ru species needs to be reduced. Such a practice is
quite common--metal halides have been widely studied as initial catalyst
sources. The reduction step is typically done in H2 flow at elevated
temperatures. For this study, however, RuCl3 did not prove to be a
successful precursor. The temperatures required to fully reduce the
metal species (500-6000 C) exceeded the pyrolysis temperature of PPAN
and caused further structural changes in the support.
In light of the difficulties associated with RuC13, Ru3(CO)12 was
selected as the metal dopant to be used in all the CO and CO2
hydrogenation studies. There are several additional advantages to the
choice of this species. The fact that the carbonyl functionality can be
easily observed in the IR makes it very easy to verify that the cluster
has initially been supported. Since both PPAN and AX21 are black, a
simple visual inspection, such as can be done with A1203 or SiO2
supports, (which are bright yellow when Ru3(CO)12 is supported on them),
can not be used to insure the success of the solvent deposition.
Another advantage to the use of Ru3(CO)12 has been shown by recent
studies which suggest that the carbonyl precursor is superior to metal
halide species because the breakdown of the cluster leads to smaller
metal crystallite sizes.151 The only disadvantage to the use of
Ru3(CO)12 is that one must be sure that the observed hydrocarbon
products are not solely from the hydrogenation of the carbonyl groups of
the cluster. To preclude any confusion, the supported Ru catalysts are
subjected to a H2 pre-treatment at reaction conditions for several
hours. During this time methane is observed as a product from the
reduction of the cluster. When this reaction subsides, the CO/H2
reactant stream is added and the Fischer-Tropsch activity videe infra)
is immediately observed. It has been shown in subsequent studies that
the long term activity and overall product distribution are the same
whether or not the H2 pre-treatment step is done.