You are on page 1of 468


Conversions of
Synthesis Gas and
Alcohols to Chemicals
Conversions of
Synthesis Gas and
Alcohols to Chemicals

Edited by
Richard G. Herman
Lehigh University
Bethlehem, Pennsylvania

Plenum Press. New York and London

Library of Congress Cataloging in Publication Data

Symposium on Catalytic Conversions of Synthesis Gas and Alcohols to Chemicals (1983:

White Haven, Pa.)
Catalytic conversions of synthesis gas and alcohols to chemicals.

"Proceedings of a Symposium on Catalytic Conversions of Synthesis Gas and Alcohols

to Chemicals, held April 6-8, 1983, at the Middle Atlantic Regional Meeting of the
American Chemical Society at Pocono Hershey Resort, White Haven,
Pennsylvania"-CIP verso t.p.
Includes bibliographical references and index.
1. Catalysis-Congresses. 2. Synthesis gas-Congresses. 3. Alcohols-Congresses. 4.
Chemicals-Congresses. I. Herman, Richard G. II. American Chemical Society. Middle
Atlantic Regional Meeting (17th: 1983: White Haven, Pa.) Ill. Title.
TP156.C35S96 1983 661'.805 83-26884

ISBN-13: 978-1-4612-9696-6 e-ISBN-13: 978-1-4613-2737-0

DOl: 10.1007/978-1-4613-2737-0

Proceedings of a Symposium on Catalytic Conversions of Synthesis

Gas and Alcohols to Chemicals, held April 6-8, 1983, at the Middle
Atlantic Regional meeting of the American Chemical Society, in the
Pocono Hershey Resort, White Haven, Pennsylvania

© 1984 Plenum Press, New York

Softcover reprint of the hardcover 1st edition 1984

A Division of Plenum Publishing Corporation

233 Spring Street, New York, N.Y. 10013

All rights reserved

No part of this book may be reproduced, stored in a retrieval system, or transmitted,

in any form or by any means, electronic, mechanical, photocopying, microfilming,
recording, or otherwise, without written permission from the Publisher

Dedicated to my scientific mentors, Abraham Clearfield, Sten

Ahrland, Jack H. Lunsford, and Kamil Klier, who, with patience, per-
severance, and example, provide continual encouragement, to my in-
spirational academic great-grandfather, Linus Pauling, to an inspi-
rational great-grandfather of catalysis, Paul H. Emmett, and to the
past and present faculty, especially of the Chemistry Department, at
Fredonia State University College, who teach the heart and mind that,
with study, effort, and optimism, nearly all goals are attainable.

Most of the papers contained in this volume are based on pres-

entations made at the symposium on Catalytic Conversions of Synthesis
Gas and Alcohols to Chemicals, which was held at the 17th Middle At-
lantic Regional Meeting of the American Chemical Society, April 6-8,
1983, in the setting of the Pocono Hershey Resort, White Haven, PA.
I thank Dr. Ned D. Heindel, General Chairman, and Dr. Natalie Foster,
Program Chairman, both of Lehigh University, for the invitation to
organize the symposium. Financial support was received from Air
Products and Chemicals, Inc. for the organization of the symposium,
and acknowledgement is made to Air Products and Chemicals, Inc. and
to the Donors of the Petroleum Research Fund, administered by the
American Chemical Society, for partial support of the conduct of the

The theme of this volume is the recent progress made in devel-

oping and understanding viable catalytic syntheses of chemicals di-
rectly from synthesis gas (CO + H2) or indirectly via alcohols. An
aim of the symposium and of this volume is to provide a meaningful
blend of applied and basic science and of the chemistry and engineer-
ing of processes that are, or hold promise to be, economically and
industrially feasible. The topics demonstrate the increasing impor-
tance of synthesis gas as a versatile feedstock and emphasize the
central role that alcohols, such as methanol, can playas chemical
intermediates. Although recent developments in, and new perspectives
of, established processes are presented, the emphasis is to provide
insights into processes that are still in the research, development,
and scale-up stages.

The practical orientation of this volume is directed towards

professional chemists and engineers. However, the papers are written
in an instructional fashion so that this volume can be used as a com-
plementary reference book in advanced undergraduate or graduate
courses in catalysis.

I wish to thank Dr. Henry Leidheiser, Jr., Director of the Canter

for Surface and Coatings Research at Lehigh University from 1968 to
1983, for his continuous encouragement and support. Appreciation is


extended to all of the authors for their pleasant and timely cooper-
ation and for their efforts in producing quality papers. I espe-
cially appreciate and thank my wife, Helen Lynn, for.her assistance
in the formatting of this volume and for typing the final copy.

Richard G. Herman
September 1, 1983


Perspectives on the United States Feedstocks for the

Production of Energy and Chemicals • • 3
Richard G. Herman

The Production of Synthesis Gas from Methane, Coal

and Biomass . • • • . • . • . . • • . . 37
Michael S. Graboski


Effects of Cobalt on Synthesis Gas Reactions over

Copper-based Catalysts • • • • • • 53
F. N. Lin and F. Pennella

Thorium-Copper Intermetallic Catalysts for the

Synthesis of Methanol 65
G. B. Atkinson, E. G. Baglin, L. J. Nicks, and
D. J. Bauer

Alcohol/Ester Fuels from Synthesis Gas • • • • • 81

John F. Knifton, Robert A. Grigsby, Jr., and
Sheldon Herbstman

Untangling the Water Gas Shift from Fischer-Tropsch:

A Gordian Knot? •••••••••••• 97
Cheryl K. Rofer-DePoorter

Some Aspects of the Slurry Phase Fischer-Tropsch Process 129

J. V. Bauer, B. W. Brian, S. A. Butter, P. N. Dyer,
R. L. Parsons, and R. Pierantozzi


Metal-Zeolite Catalysts for the Conversion of Synthesis

Gas to Selected Hydrocarbon Products • • • • • • • 151
v. U. S. Rao, R. J. Gormley, R. R. Schehl, K. H. Rhee,
R. D. H. Chi, and G. Pantages

Conversion of Synthesis Gas to Olefins over Physical

Mixtures of High Si02/A1203 ZSM-5 and Fe(K) • • • • • • 167
F. G. Dwyer and W. E. Garwood

Catalyst Support Effects on Selectivity in the

Fischer-Tropsch Synthesis 179
J. G. Goodwin, Jr., Y. W. Chen, and S. C. Chuang


The Use of Perfluoroalkanesulfonic Acids in the

Palladium-Catalyzed Carbomethoxylation of Olefins 193
F. J. Waller

Phosphine Modified Cobalt Carbonyl Catalysts for the

Hydroformylation of Dicyclopentadiene 203
Clayton D. Wood and Philip E. Garrou

Ethylene Glycol from Methanol and Synthesis Gas via

Glycolic Acid • • • • • • • • • • • • • 221
S. Suzuki, J. B. Wilkes, R. G. Wall, and S. J. Lapporte

SYfithesis Gas to Formic Acid via Methanol Carbonylation 249

Alan Peltzman

Recent Advances in Alcohol Homologation: The Effect of

Promoters 261
W. R. Pretzer and M. M. Habib


Polyethers and Organorhodiums: A Study of Oxidative

Addition and Transfer Hydrogenation • • • • • 287
M. L. Deem

Synthesis of High Octane Ethers from Methanol and

Iso-Olefins • • • • • • • • • • • • • 307
Jack D. Chase

Conversion of Methanol to Low Molecular Weight Olefins

with Heterogeneous Catalysts • • • • • • • • • • • 323
Lujia Liu, Ricardo Garza Tobias, Kenneth McLaughlin,
and Rayford G. Anthony

Catalytic Conversion of Alcohols to Olefins 361

Oemer M. Kut, Robert D. Tanner, J. E. Prenosil, and
Kenneth Kamholz

Conversion of Methanol to Hydrocarbons on Heteropoly

Compounds • • • • 395
J. B. Moffat and H. Hayashi

Formaldehyde from Methanol • • • • • • • • • • • • • 413

C. J. Machiels, U. Chowdhry, R. H. Staley, F. Ohuchi,
and A. W. Sleight

Catalytic Conversions of Methanol to Chloromethanes 419

S. Akiyama, T. Hisamoto, T. Takada, and S. Mochizuki

Alkylation of N- and O-Heteroatom Compounds with Alcohols,

with Special Reference to the Synthesis of Alkylamines. 433
Richard G. Herman


1. U. S. Energy Conversion Factors 4~

2. Chemical Nomenclature 465

INDEX ............................ 469



Richard G. Herman

Center for Surface and Coatings Research

Sinclair Laboratory, #7
Lehigh University
Bethlehem, PA 18015


We tend to think of energy as power that is consumed to provide

us with services, e.g. electricity for lighting, fuel oil for heat,
and gasoline for transportation, and of chemical feedstocks as raw
materials that are transformed into industrial or consumer goods.
However, energy and chemical feedstocks are not exclusive categories
because both are based principally on the same natural resources.
Coal, crude oil, and natural gas provide raw materials for industry,
as well as the energy to process the materials into commodities. It
will become clear in the following discussion that energy, industry,
consumer goods, and the standard of living are intimately entwined.


To carry out its daily functions, the human body consumes about
3000 kcal, or approximately 12,300 Btu, of energy. Of course, we
enjoy eating and drinking to provide this energy, wherein our bodies
transform the food into energy. This quantity of energy is equiva-
lent to the energy contained in one lb of bituminous coal. However,
to maintain our standard of living in the United States today, the
per capita energy consumption is equivalent to about 12 short tons
of coal annually [1]. This 300 million or so Btu quantity of energy
is used for transportation, to run our households, to operate our
business establishments, and by our industrial sector. It has been
noted that this per capita energy usage is equivalent to each of us
having 1200 personal slaves [2].

From where is this energy derived? It is obtained mainly from
the storehouse of fossil fuels - coal, crude oil, and natural gas.
Figure 1 shows the apportionment of natural resources that are used
to generate energy in the United States. A similar figure could be
constructed to represent the world energy patterns [3], although the
early usage of coal would be at a larger percent of the total, while
the contribution of petroleum would occur somewhat later and would
be of a slightly smaller magnitude. It is clear from Figure 1 that
the source of U.S. energy shifted from wood to coal in the late
1800's, most noticeably in the 1880-1895 period, and shifted again
after World War II from coal to petroleum and natural gas. The con-
tribution of nuclear power has been steadily increasing, while hydro-
derived power has maintained a constant portion of the energy supply.

Nuclear and Hydroelectric Power

Subsequent discussion will center on coal, petroleum, and nat-

ural gas since these resources can be used to produce chemicals, as
well as energy. Nuclear power and hydroelectric power are devoted
to the generation of electricity, and the role of these in electrity
generation will be briefly discussed to provide a perspective on the
part that these play in the balancing of energy resources. The
annual production of consumable electric energy in the U.S. has sta-
bilized at about 2,300 billion kilowatt hour (kWh), which is equiv-
alent to 11.2 x 1015 Btu or 11.2 Quads. To generate this electricity,
large quantities of fuels are consumed in a rather inefficient manner
because of conversion and transmission losses. In fact, it requires
about 10,000 Btu's to deliver one kWh of electricity [4]. This cor-
responds to a 60-67% loss of the energy contained in the original
fuel. Therefore, it now takes about 3.3 Quads of nuclear energy and
3.3 Quads of hydroelectric power, as well as on the order of 13 Quads
derived from coal, 4 Quads produced from petroleum, and 4.5 Quads
obtained from natural gas to generate the quantity of electricity
annually utilized in the U.S. Thus, appreciable consumption of fossil
fuels is necessary to satisfy the demand for electricity.

This distribution of fuels to generate electricity is the aver-

age for the U.S. Some states, e.g. Illinois, depend heavily on nu-
clear power generated electricity, while others, such as those asso-
ciated with the Tennessee Valley Authority, rely on hydroelectric
power. Still others, such as the states bordering on the Ohio Valley
derive electricity from coal-fired power plants. This leads to dif-
ferences in the cost of electricity to the consumers in the various
areas of the U.S. An estimate of comparative costs is given in Table
I, and it is indicated that electricity can be produced from coal-
fired and nuclear power plants for about the same cost of 38-39
mills/Kwh. This assumes that coal can be obtained at $40/ton and
that the nuclear plant construction lead time is maintained at only
six years. If the price of coal rises to $65/ton, the fuel cost in-
creases to 26.0 mills/Kwh, which results in a total cost for the

z 75
0 50

z 25
D.. 0
1850 1875 1900 1925 1950 1975 2000
Figure 1. The historical pattern of energy utilization in the
United States.

generated electricity of 48.2 mills/Kwh [5]. If the construction

lead time for the nuclear power plant is drawn out to 10 years, the
total cost of the produced electricity will increase by approximate-
ly 25% to about 49 mills/Kwh.

During the last decade, the price of coal has been increasing
in conjunction with the increased cost of crude oil, although not as
rapidly. This has led to even more favorable comparative economics
for utilizing nuclear power to generate electricity in Western Europe
and Japan. This point was initially emphasized by the very large
price increase for crude oil that was enforced by the OPECa nations
in 1973. During the following year, there were 62 nuclear power re-
actors ordered or letters of intent placed by the OECDb countries.
Of course, with a six year lead time, these plants would not begin
to come on stream until 1980. Between 1974 and 1980, however, new
governmental regulations, increased lead time and costs, and, due in
part to conservation, the average annual increase in electricity
demand by OECD nations dropped from the high demand rate of about 7%
annually in the 1960-1973 period to 3.1% [6]. This led to the defer-
ment or cancellation of 71 nuclear power projects in the U.S. during
1974-1981 [6]. This compares with the 72 operating nuclear power
reactors at the end of 1978 [7]. The trend in the U.S. for new elec-

aOrganization of Petroleum Exporting Countries (OPEC): Algeria, Ec-

uador, Gabon, Indonesia, Iraq, Iran, Kuwait, Libya, Nigeria, Qatar,
Saudi Arabia, the United Arab Emirates, and Venezuela.
bOrganization for Economic Cooperation and Development (OECD): Aus-
tralia, Austria, Belgium, Canada, Denmark, Finland, France, Germany,
Greece, Iceland, Ireland, Italy, Japan, Luxembourg, Netherlands,
New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, Turkey,
United Kingdom, and the United States.

Table 1. Comparative Cost Estimates (1981:mills/Kwh) for Electric-

ity Generated from High Sulfur Fuel Oil, Nuclear Power,
and Coal [adapted from Reference 5].

OH a Coal a Nuclearb

Plant size (megawatt) 600 600 1100

Capital cost 12.9 17.1 24.8
Operating cost C 4.2 5.1 4.2
Fuel cost 47.6 16.0 10.0
TOTAL COST 64.7 38.2 39.0

Capital investment ($/Kw) 692 920 1331

Construction lead time (yr) 3 4 6

Fuel cost $27/bbl $40/ton

Relative cost d 194 60 40
Conversion efficiency(%) 35 60 40

aWith flue gas desulfurization

bpressurized water reactor
cCapacity factor of 65% for 30 years
d$ per ton oil equivalent

trical capacity is now away from nuclear power and toward coal-fired
generation plants. For example, from mid 1983 to 1988, 50 nuclear
plants are scheduled to start up, as well as 52 new coal-fired in-
stallations [8]. Past that date, 11 nuclear power plants are still
on the drawing boards, while 72 coal-based plants are in the plan-
ning stage [8]. .

This situation is of special concern when considering fuels uti-

lization, developments in the mining of coal, distribution of elec-
trical power, and possible safety and environmental problems connect-
ed with the coal indistry. Of more immediate concern is the fiscal
stability of the capital intensive nuclear power industry. As of
December 31, 1981, the 76 U.S. nuclear power reactors (not plants)
were located in 27 states and had a 74,000,000 kW operable capacity
[9]. Most of these were associated with public utility systems that
were or had been constructing or had planned additional nuclear re-
actors. Many of these additions were abandoned but were still a fi-
nancial liability because of the debt incurred in these projects.
This culminated in August 1983 when t~ Chemical Bank of New York
filed suit against the Washington Public Power Supply System (WPPSS)
seeking to recover q $2.25 billion debt from the largest bond default
in U.S. history flO]. The WPPSS had developed a $24 billion con-

struction program to build five nuclear power plants, and one of

those will begin operation in 1984. It should be noted that 13 of
the 36 largest hydroelectric plants, with a combined installed capac-
ity of 39,790,000 kW for the 36 plants, in the "U.S. are located in
the state of Washington [9J.

Fossil Fuels and Economics

From this discussion, it can be concluded that nuclear energy

will not appreciably increase its percentage of the total energy
consumed in the U.S. in the near future (Figure 1), and it will not
alleviate the demand for fossil fuels. Figure 2 depicts the trend
of energy consumption in the U.S. for the last 33 years. It is evi-
dent that since the Middle East disruption of 1973, the consumption
of energy has stabilized in the 70-80 Quad range, where 1 Quad = 1
quadrillion Btu. The energy produced in this country, however, does
not fulfill the demand, and Figure 2 shows that since 1957, the U.S.
has been an energy importing nation. Figure 2 also shows that petro-
leum is the dominant fuel in satisfying the U.S. energy demand. Al-
though the U.S. continued to export coal, mainly to Europe and Japan,
the importation of petroleum steadily increased until 1977 when about
45% of the domestic oil demand was met by imported crude oil. This
resulted in a significant drain on the U.S. economy because fully
25% of the 76.3 Quads of energy consumed was derived from imported
petroleum, which resulted in a massive cash flow from the U.S.

At the same time that U.S. crude oil imports were increasing,
the price was also increasing. This is shown in Figure 3, where a
barrel of petroleum contains 42 U.S. gallons. The most immediate and
personal impact of the price increase of crude oil was the accompa-
nying increase in the price of gasoline. It can be noted from Figure
3 that the % increase in the price of a gallon of gasoline was not
even near to the % increase in the cost of a barrel of gasoline.
Part of this is due to the taxation contained in the price of gaso-
line, e.g. in 1973, 38% of the real price of a gallon was added as
tax [llJ. Thus, the "real" cost of gasoline in 1973 was approximate-
ly $.27/gallon. In 1980, the % tax in gasoline prices was only 13%
in the U.S., that is a real price of $1.06 plus $.14 tax for a sell-
ing price of $1.20/gallon. In 1973, gasoline prices in Europe, and
most other countries of the world, were already over $l/gallon be-
cause of taxation. For example, in Germany 239% of the "real" cost
of gasoline was added to the price as tax. Even so, the price of
gasoline only increased about 3.5-fold in the 1973-1980 period, while
the cost of imported crude oil increased over 10-fold.

On the other hand, industrial oil prices increased at double

the rate of the gasoline price increase. One reason for this was
that drivers could easily refuse to consume as many gallons of gaso-
line as they might otherwise have done, and this was particularly
true in the U.S. as compared to most other OEeD countries. However,

U. S. Consumption I' /-\_,~
70 ~/
2 I
co 60 1
.;: 50 I
I ·
-:;: I
0' 40 ./....,'\)
./ _'-''-.1
~ 30
z 20
Petroleum ~


Figure 2. Comparison of the production ( ••• ) and consumption (---)

of energy in the United States shows that the U.S. is a
net importer of energy. The energy produced from imported
petroleum (_) can be compared with the total quantity of
energy derived from petroleum (0).

utilities and industries could not decrease their consumption of pe-

troleum products without decreasing their goods and services. They
basically comprised a captive market because they could not readily
apply interfuel substitution. The same was true of the U.S. residen-
tial sector of the economy.

A barrel of crude oil is refined into a variety of products,

and, contrary to what might be believed, an incoming tanker of crude
oil cannot be converted into an outgoing tanker of gasoline. Typi-
cally, only 15% of a barrel of crude oil is refined into gasoline.
Other fractions include 37.4% middle distillates consisting of domes-
tic heating fuel oil, diesel fuel, jet fuel, and lubricants, 31% in-
dustrial boiler fuel (high sulfur fuel oil), 3.2% liquified petroleum
gas (LPG), 0.8% asphalt and tar products, 6.5% naphtha, and 6.1% re-
finery fuel and losses. Most of these fractions suffered price in-
creases that were greater than those for gasoline.

Naphtha is a primary chemical feedstock, and it can be easily

thermally or catalytically cracked to olefins and other petrochemi-
cals. Of the olefins, ethylene can be converted to polyethylene,
ethylene oxide and ethylene glycol, and vinyl chloride and styrene
monomers, while propylene can be transformed into polypropylene,
acrylonitrile, isopropanol, propylene oxide, and many other basic
chemicals. Naphtha can also be reformed into the xylenes, toluene,
and benzene. It can also be catalytically reformed into synthesis
gas. Thus, naphtha is a cornerstone of the U.S. chemical industry.


~ 30
Co 25

6 20
:::l 2 =g
a: a:
() 15 a:

f0- 1.40 ~ ~
1.20 ~ tit
10 1.00
~ ~
UJ w 0
0 .80 ~ ~
0 a: "
~ 5
.. ....
~ ~
...I ~

.20 '" "

~ ~
;:: a:
o Z'" 0
1960 1970 1980

Figure 3. Plots showing the cost of Middle East crude oil and of
gasoline in the United States during the last 33 years.

With the exceptions of ammonia, urea, methanol, and formalde-

hyde, which are principally derived from natural gas, virtually all
commodity organic chemicals are derived from petroleum. The disrup-
tions and price increases of the last decade have brought the real-
ization that the world's oil supply is not unlimited and that coun-
tries that are not energy su1f-sufficient are economically vulnerable
to outside forces. This has led to political decisions that strive
toward energy independence by encouraging domestic exploration for
new energy reserves, new energy resources, and new methods of trans-
forming energy from one form into another.


In 1981, the U.S. produced 3.1 billion barrels of crude oil,

which placed it third behind the USSR (4.5 billion barrels) and Saudi
Arabia (3.6 billion barrels) in total production. However, these
rates of production cannot be long maintained, as indicated in Table
2, unless new major oil fields are found. The adapted data [12] in
this table are given for the ten countries that possess the largest
proved reserves. The data for the USSR and, in particular, for China
(PRC) might not be very accurate, and, in fact, the proved petroleum
reserves for China might be 2-3 times greater than the stated quan-
tity. The years of supply given in Table 2 are obtained by assuming
that no new reserves are found. Of course, there is continuous

Table 2. 1981 Proved Reserves, Actual Production, and Number of

Years Supply at the Current. Rate of Withdraw from Reserves
(million barrels)

Country Proved Petroleum Years of

Reserves Supply

Saudi Arabia 168,026 3,600 46.7

Kuwait 67,934 401 169.4
USSR 63,001 4,500 14.0
Iran 57,497 475 121.0
Mexico 44,002 840 52.4
Iraq 30,002 330 90.9
Abu Dhabi 28,997 550 52.7
United States 26,403 3,100 8.5
Libya 23,001 435 52.9
China 20,502 770 26.6

drilling activity and new reserves are being found. However, much
of the world has been extensively explored, and the probability of
finding additional super-giant oil fields (initial reserves greater
than 5 billion barrels) is dwindling. Indeed, during the 1971-1979
period, only two super-giant fields have been discovered, and both
of them are assigned to Mexico (with combined recoverable reserves
of 15 billion barrels) [13]. Of the 35 known super-giant fields,
25 are in the Middle East and 10 of those are in Saudi Arabia. One
of the remaining super-giant fields is the Alaskan Prudhoe Bay field.
It can be noted that during the 1971-1979 period, 40 giant fields
(initial reserves between 0.5 and 5 billion barrels) were discovered,
but this is only 58% of the number discovered during the previous
decade [13]. It is of interest that the 35 super-giant fields and
the 245 giant fields together account for 75% of the current produc-
tion and 73% of the proved reserves. The 29,700 smaller oil fields
provide the remainder [13].

The Oil and Gas Journal provides running statistical analyses

of drilling activity and World Oil periodically updates the drilling
activity around the world. During the last decade, as well as his-
torically, 90% of the wells drilled have been in the U.S. This has
resulted in about 550,000 oil wells in the U.S. The average well
produces about 16 barrels/day, which can be compared to the average
well in Saudi Arabia that produces about 35,000 barrels/day. The low
U.S. number results from the fact that two-thirds of the U.S. wells
average less than 3 barrels of output per day and contribute only 13%
to the total production. A summary of the annual new found oil and
the annual production, partially obtained from reference [14], is

20 Proved Reserves
(f) (f)
...J ...J
w w
a: 16 a:
a: 30 a:
« «
m 12 m
0 20 0
(f) 8 (f)
z z
Q 10 Q
...J 4 ...J
...J ...J
m m
1950 1955 1960 1965 1970 1975 1980 1985

Figure 4. Proved reserves of crude oil at end-of-year in the United

States. Production (annual withdrawals from reserves)
has generally exceeded the additions of new proved re-
serves during the last two decades. In 1970, the Alaskan
Prudhoe Bay reserves were added.

given in Figure 4. It is evident that since 1960 production of oil

has been outstripping the addition of new reserves, except for the
years 1970 and 1981. In 1970, the Prudhoe Bay reserves were added,
and the 1981 additions were due to enhanced drilling activity (up
45% from 1978 and 1979) and off-shore finds.

The alaskan oil fields are a significant portion of the U.S.

reserves and produce a significant portion of the petroleum used in
the U.S. This is shown in Table 3. It is probable that production
from the lower 48 states will gradually decrease. It is hoped that
relatively unexplored Alaskan areas, such as the Beaufort Sea, Lis-
burne, the Bristol basin, the Navarin basin, and the Outer Continen-
tal Shelf will yield substantial oil fields. All of these appear to
be high potential areas that could provide petroleum to the U.S. for
a number of decades. Although the trend for imported petroleum is
certainly downward, it is not known when and where it will stabilize.
It is evident from Table 3 that part of the decline in import oil is
due to less crude oil being consumed in the U.S.

Natural Gas

Natural gas is used as a feedstock for certain bulk commodity

chemicals, the most important of which are methanol and ferti1izer-
grade ammonia. The latter two chemicals account for 9% of all nat-
ural gas used for industrial purposes [15]. The U.S. Natural Gas
Policy Act (NGPA) of 1978, which went into effect in December 1979,
decontrolled the price of gas produced from wells drilled after April
20, 1977. As more of the production of natural gas is derived from
the new wells, the average price of natural gas will increase. Al-
though the average price of natural gas delivered to steam and elec-
tric generating plants was 28.3% lower than residual oil (but 105.7%

Table 3. Source of Crude Oil used in the United States (million


Year Lower 48 States Alaska Imported

1976 7.95 0.17 5.30

1977 7.75 0.50 6.59
1978 7.48 1.23a 6.19
1979 7.18 1.37 6.44
1980 6.95 1.65 5.23
1981 6.93 1.64 4.14
1982 6.93 1.73b 3.34

aprudhoe Bay field began significant production

bKuparuk River field began producing

higher than coal) [16], the price increases have already been felt.
For example, the % gas cost outlay made by gas utilities to natural
gas producers more than doubled during the 1972-1982 period [17], and
this has been reflected in the frequent requests of gas rate increas-
es made by gas utility companies to state public utility commissions.
The increased price of natural gas is also being felt in the chemi-
cal industry. In fact, it has been estimated that by 1985, feedstock
natural gas will represent more than 85% of the cash cost of produc-
ing methanol [15].

With this in mind, what are the prospects for the future produc-
tion of natural gas in the U.S.? The U.S. is the largest gas produc-
ing nation in the world, but Figure 5 shows that the production
has been gradually declining since the 1970-1973 period of high pro-
duction. In light of the crude oil situation during the last decade,
this would hardly have been predicted. However, the low volume of
additions during the 1968-1978 period could perhaps have been pre-
dicted because of the lack of incentives for drilling and finding new
sources of natural gas. The NGPA of 1978 provided the basis for eco-
nomic incentive, and the finding of new gas has increased.

Although the U.S. is the largest consumer of natural gas (>40%

more than second place USSR and over 10-fold more than third place
West Germany), there is currently a surplus of natural gas in the
U.S., due in part to conservation measures taken by industrial and
residential sectors. It is expected that the surplus will disappear
before complete deregulation takes effect in 1985. As the average
price of natural gas continues to rise, drilling activity will remain
high, but the finding rate per foot drilled will continue its steady

Proved Reserves I-
LL 30 240
~ ~
~ ..-Production ~

U 20 160 U
a a
~ 10 80 z
a 2...J
...J :::
a: a:
I- 1950 1955 1960 1965 1970 1975 1980 1985 I-

Figure 5. Proved reserves of natural gas at end-of-year in the

United States. The Prudhoe Bay reserves were added in
1970. Production of natural gas peaked during the 1970-
73 period.

Figure 5 indicates that the U.S. is annually discovering as

much natural gas as it is consuming, and that a 10 year supply is
being maintained as proved reserves. The current production is de-
rived from the lower 48 states, which contains 85% of the proved re-
serves. Natural gas from Alaska will make a significant contribution
in the near future, and it is likely that new and extensive gas
fields will be discovered in Alaska. However, there is no consensus
on the probable size of these fields, nor on the prospects of pro-
duction from new deep (>15,000 ft) gas discoveries. Extraction of
natural gas from tight sands might make a significant contribution
by the year 2000, while general use of synthetic natural gas by that
time is not expected.


The world contains vast reserves of coal, and, similar to petro-

leum, the reserves are not uniformly geographically distributed.
Most of the coal reserves are located in the northern hemisphere, and
over 60% of them are found in the United States, the USSR, and China
(PRC). Table 4 presents one estimate of the world's recoverable
proved reserves of coal [18]. Other estimates differ for some coun-
tries, e.g. India might only have 12 billion tons of recoverable coal,
while it has been estimated that Poland has 60 billion tons [19].
The United States might have 181.9 billion tons of recoverable coal
[20] or even as much as 200-260 billion tons that can be recovered
[21], e.g. 250 billion tons [22]. Even more optimistic estimates
placed the recoverable coal reserves at about 430 billion tons [7,23].
The quantity of coal referred to here as recoverable is approximately
half of the proved reserves in the U.S. because recoverable coal is
considered to be mined profitably from coal seams at least 24" thick
using current technology. It can be noted that the U.S. geological
coal resources are thought to lay in the 2570 [18,19] to 2924 [20]
billion short ton range, consistent with the usual rule-of-thumb

Table 4. An Estimate of the World's Technically and

Economically Recoverable Reserves of Coal

Country Recoverable Reserves

(billion short tons)

United States 177

USSR 110
China (PRC) 99
United Kingdom 45
Germany 35
India 33
Republic of South Africa ~7
Australia 27
Poland 21
Canada 10
Others 53

estimation that the quantity of geological resources such as fossil

fuels and minerals are at least 10-20 times greater than the proved

The United States is the largest coal producing nation in the

world, mining nearly 30% of the world's total. China and the USSR
are the other two countries that mine large quantities of coal, each
producing about 20% of the total. The trends in U.S. coal production
are shown in Figure 6. It is evident that anthracite coal, which is
mainly obtained from deep mining in Pennsylvania, no longer has a
significant impact on the production of energy. In fact, only 89
deep anthracite mines are still in operation [24].

Of the coal produced in the U.S. in 1981, 67% went to electric

utilities [25]. As previously noted, coal usage by this sector will
increase due to new coal-fired generation plants that are being built.
In addition, it is estimated that two-thirds of the existing 148 GW
oil-based power plants'will be either phased out or converted to coal
plants in the next ten years [26]. Of the remainder of the coal pro-
duced in 1981, 15% was exported, 17% was consumed by industry, and
<1% was utilized by the residential and commercial sectors. Of the
17% consumed by industry, about half was metallurgical coal for coke

In 1977, the U.S. production of coal was 689 million tons, and
President Carter released an energy plan that would increase the
coal production to 1,265 million tons in 1985. Of this production,

.8 700

c 600
;i 500
0 400
<t 300
:::> 200

1850 1900 1950 2000

Figure 6. The historical production of coal in the United States.
The production of anthracite in Pennsylvania, which has
97% of the anthracite reserves has declined to a very
low level.

7% would be exported while 62% would go to the electric utilities,

22% would be consumed by industry, 8% would be utilized for metal-
lurgical purposes, and 1% would be converted to synthetics [7].
Considering Figure 6, it is doubtful that coal production will ap-
proach the goal set in 1977. However, it should be noted that the
production of synthetic chemicals from coal will be initiated in the
U.S. during late 1983 when the Tennessee-Eastman plant is placed on
stream. The plant, located in Kingsport, TN, is the first of the
large modern, new-generation installations to produce commodity chem-
icals [27], and it will produce acetic anhydride from synthesis gas
and methanol obtained from coal.

Although synthetic fuels and chemicals are not expected to have

much of an impact on coal production or consumption during the 1980s,
they could effect a pronounced demand on coal during the 1990s. How-
ever, during the 1980-1987 period, the U.S. government, via the Syn-
thetic Fuels Corp., is supporting $15 billion of research and devel-

opment with the goal of producing 500,000 barrels of oil equivalent

per day of synthetic fuels from coal by 1987. A relatively recent
report estimated that the production of synthetic fuels from coal
could reach the 1.9-4.9 million barrels of oil equivalent per day
(Mbdoe) level by the year 2000 by the combined OECD countries [28].
Conversions of this scale would require 300-800 million tons of coal
annually. If this were carried out with U.S. coal, it would require
increasing the current output of coal by 38-100%. The study assumed
that the commercial synthetic fuel plants would produce 50,000 bar-
rels/day. Therefore, the lower estimates are more likely than the
higher estimates. However, it should be noted that synthetics could
be made from synthesis gas produced by underground gasification of
coal, which would by-pass the actual mining of the coal.

Usage, Conservation, and Efficiency

In the late 1960s, the U.S. economy was strong, inflation and
unemployment were at low rates, and there was a pronounced movement
away from coal and toward petroleum. Industrial capacity was being
utilized at record levels. However, in the early 1970s the economy
was slowing down and commodity prices were increasing. The increase
in OPEC petroleum prices and crude oil disruption of 1973 deepened
the economic slowdown and extended it into 1974 and 1975. Histori-
cally, there is a close direct relationship between economic activity
and energy consumption. Table 5 shows that this is true during the
1976-1979 recovery period. During this four-year period, the GNP
increased at an annual rate of 4.7% and energy consumption increased
at an annual rate of 2.8%. Adjustments were gradually being made to
alleviate the heavy U.S. dependence on imported petroleum and the
Natural Gas Policy Act was finally passed in late 1978, which encour-
age the drilling for and production of natural gas. Direct and in-
direct coal liquefaction research and development programs were
strongly supported by DOE and private industry, although the growth
of this support had taken a period of time to develop.

The drastic price increases for petroleum (Figure 3) imposed by

OPEC in 1979-1980 came as a shock. The immediate results were a
recession with inflation and unemployment rates that rose above 10%~
Fiscal and monetary policies were invoked that tightened the economy,
but led to decreased importation of petroleum (Table 3) and suppres-
sion of inflation. The GNP showed a 0% growth rate from 1979 to
1982 (in 1972 dollars), while the total energy consumption over that
period fell by 9.0%, as indicated by Table 5. The cost to the U.S.
economy is still not known since recovery is just beginning to show
in 1983. However, if the 4.7% growth rate in GNP had been maintained
from 1979 to the present, the 1983 GNP would have been $1,777 bil-
lion (in 1972 dollars). This indicates that the 1979-1980 OPEC oil
prices have to-date cost the U.S. over one-quarter trillion dollars
in real income!
Table 5. U.S. Energy Efficiency m
GNP (billion) Energy Consumption 103 Btu per % Increase in -<
Year »
1972 dollars (trillion Btu) GNP dollar Energy Efficiency z
1974 1,246 72,759 m
58.4 s:
1975 1,232 70,707 57.4 1.7 n
1976 1,292 74,510 57.4 0.0 »
1977 1,370 76,332 55.7 3.0 en
1978 1,439 78,175 54.3 2.5
1979 1,479 78,910 53.4 1.7
i980 1,474 75,913 51.5 3.6
1981 1,503 74,123 49.3 4.3
1982 1,476 71,800 48.6 1.4
1983a 1,518 72,870 48.0 1.2

aEstimated quantities


One outcome of the last decade of petroleum price escalations

has been a significant increase in the efficient utilization of
energy in the United States. This is demonstrated in Table 5, where
it can be seen that a 17.8% drop has occurred during the last decade
in the number of Btu's required to generate $1 of GNP. This increase
in more efficient energy usage is reflected in every sector of the
economy (Table 6). The decreased rates of growth for the 1974-1980
period are mainly a reflection of the 1980 energy consumption, but
the energy usage has been even less than the 1980 consumption for
every year since then. It is evident from Table 6 that energy con-
servation in the home and in the business place has been significant.

Forecasts and Actual Usage of Energy

Forecasts of economic activity are continuously made to aid in

investment planning, introduction of n.ew product lines, enlargement
of plant capacity, etc. An example of an estimate made in 1975 of
the U.S. consumption of energy in 1980 and 1985 is given in Table 7.
This, of course, was made after the first OPEC crude oil disruption
but without knowledge of the forth-coming 1979-1980 OPEC price in-
creases. Comparison of the 1980 figures shows that the projected
energy consumption was about 15% higher than the actual U.S. usage.
This was primarily caused by a lower consumption of crude oil than
was predicted in 1975. However, the consumption of coal to generate
energy was also lower than planned, and the utilization of nuclear
power to generate electricity was only about 60% of that expected.
It appears that the estimated 1985 total usage will differ from the
actual usage by about 25%. In the long-term, this energy "savings"
will benefit the U.S., especially since this savings will be annually


The past decade has demonstrated that unpredictable events can

occur that drastically alter the uniform course of economic progress
on an international scale by influencing the availability of energy.
The intimate relationship between energy consumption and economic
growth leads to a stagnation of economic growth if total energy con-
sumption stabilizes, unless adjustments are made. The obvious ad-
justments include increases in efficiency of energy utilization and
conservation. Everyone has been faced with conservation in the use
of household energy and personal transportation. Many industrial
companies have made marked advances in the efficient use of energy
by replacing old electric motors by newly designed motors that are
much more efficient, by retrofitting commercial gas burner systems
[31], by redesigning electric furnaces [32], and by utilizing 'waste'
heat. However, conservation and increases in energy utilization ef-
ficiency are not sufficient for other than a short period of time
because of the continuous population increase and the desire to con-

Table 6. Annual Growth Rates of Energy Consumption by the Principal

U.S. Sectors(%) and Consumption (in Quads) for two Partic-
ular Years

Annual Growth Rates Consumption

1960-73 1973-81 1973 1981

Electricity generation "'-7 2.6 20.0 24.6

Transportation 4.4 0.5 18.5 19.2
Industrial & miscellaneous 3.9 -2.7 23.9 19.2
Residential & commercial 3.9 -1.6 12.2 10.7

tinuously improve the standard-of-living.

At this point, it might be appropriate to point out that, al-

though many industrial nations have made major advances in control-
ling their rates of population growth, the population of the earth
has been increasing at an exponential rate. It is estimated that
2000 years ago, the population of the earth was about 300 million
people. Somewhat over 1800 years later, the popula~ion reached the
1 billion mark. About 100 years after that (1930), the population
had doubled to 2 billion. The population reached 3 billion people
only 30 years later (1960), and then reached 4 billion people 15
years after that. These numbers provide some perspective to the huge
quantities of natural resources that must be produced and of energy
that is consumed.

With these points in mind, a schematic of the energy supply and

demand during the last decade is shown in Figure 7. The supply of
energy was abundant during the 1960's, and, in fact, the 'real' price
for petroleum slightly declined during that period. The 1973 crude
oil price increase and embargo created an energy shortage and en-
hanced the economic slowdown, which continued through 1974 and 1975.
Recovery began in 1976, and the energy consumption of 1976 was the
same as for 1973 (74 Quads), but the GNP was somewhat higher. The
drastic increases in the OPEC oil prices in 1979-1980 again created
a shortage of consumable energy, and the U.S. energy consumption
again fell to 74 Quads (Table 5). It is clear here that the supply
of energy is not necessarily only limited by physical constraints,
e.g. not enough crude oil pumped or not enough electricity generated,
but the supply can be suppressed by economic factors.

It is evident that the U.S. proved reserves, as well as those

of the world, of petroleum, natural gas, and coal are finite. In
particular, the U.S. will exhaust its proved reserves of crude oil

Table 7. U.S. Energy Usage and the Sources of the Energy (quadrillion Btu)

1980 1980a 1981 1982b 1983b 1985a

Oil 34.202 41.040 32.113 30.540 30.640 45.630

Natural gas 20.394 20.600 19.911 18.930 19.180 20.100
Coal 15.461 17.150 16.101 15.970 16.500 21.250
Hydro- and geothermal 3.184 3.800 3.097 3.330 3.330 3.850
Nuclear 2.672 4.550 2.901 3.030 3.220 11.840
Oil shale 0.870
Total usage 75.913 87.140 74.123 71.800 72.870 103.540

aEstimates made in 1975 [29].

bApparent actual usage [30].



SUPPL Y r---~


1973 1976 1979 1982

Figure 7. A schematic of the recent supply and demand for energy.

and natural gas by the end of the century if new reserves are not
continuously discovered (Figures 4 and 5) and if better and more ef-
ficient means of extraction, e.g. surfactant and C02 injection, are
not further developed. The presently recoverable reserves of coal
in the U.S. are at least 177 billion tons (Table 4), and at the cur-
rent rate of mining, these reserves would last for about 220 years.
Regardless of the number of years the reserves of fossil fuels might
last, the fact of finite quantity must be emphasized.
At some time in the future, the supply of fossil fuels will
reach a limiting point, at which the physical supply of these fuels
will decrease due to irreversible depletion of reserves. This is
shown schematically in Figure 8 on a prolonged time scale. The
demand for energy derived from fossil fuels has increased with pop-
ulation growth and with technological and social advances. The
energy demand curve in Figure 8 could also be labelled as standard of
living. As the fossil fuels become depleted, they will initially be
both physically and economically limited for energy production. Con-
servation and enhanced usage efficiencies, as well as greater depend-
ence on current alternatives, will delay the effects of the depletion
of fossil fuels. However, in the long run, the demand for energy
will fall because there will be no energy available if new energy
forms have not been developed. Hydroelectric and current-generation
nuclear power will provide little per capita energy. Ultimately,
human society will return to basically an agrarian labor-intensive
society (curve a in Figure 8).

Of course, this is a drastic scenario, which would, no doubt,

involve intensive wars for control of energy resources. In contrast,
new methods of large-scale energy production could arise. These
might involve the development of energy generation utilizing wind
power, wave power, ocean thermal differences, direct solar heating
and conversion, terrestrial thermal power, and sophisticated biomass
conversion techniques. These could provide a sustainable source of
energy and standard of living (curve b), even though many of these do
not possess high efficiencies. Development of additional large-scale

FUELS r---------__ ",-
- - - -c
/ .......................... ·······b



Figure 8. A schematic showing three scenarios for future energy

demand as the supply of fossil fuels is depleted.

sources of energy, e.g. fusion energy, could even lead to further

enhancement of the average standard of living (curve c).

An obvious question arises in regard to the time frame involved.

Figure 8 would indicate a rather long period of time might be in-
volved in this world-wide process. However, a sense of the time
frame for the U.S. will be illustrated by utilizing the consumption
of coal, its most abundant fossil fuel. A reasonable estimate of
the U.S. proved recoverable reserves of coal is 177 billion tons [18].
Beginning this year (1983), with an annual production of 800 million
tons (Figure 6), and assuming an annual growth in production of 5%
(in an attempt to decrease and compensate for imported petroleum),
Figure 9 demonstrates that the currently known coal reserves will be
completely consumed in about 50 years. If the recoverable coal re-
serves were actually double the stated quantity, coal would be pro-
duced for 14 additional years. Nearly tripling the amount of recov-
erable coal hardly increases the years of production further. This
figure emphasizes the exponential-nature of production involving a
constant annual growth, and a similar effect was noted in the discus-
sion of population growth.

As a natural resource is depleted, it is unreasonable (and un-

desirable) to expect that its production will show net gains. Table
8 contains a scheme of coal production and consumption based on the
1983 production being 800 million tons. The data in the table as-
sumes an annual growth in U.S. coal production of 5% for the next 25
years. This would lead to a more than tripling in the total annual
production of coal by the end of the 25 year period. At the end of
75 years (2058), the annual production of coal would be 10 times the
1983 production. This would appear to be a feasible goal, especially
if the coal consumed by underground gasification is included.

However, there are many factors that will inhibit the growth of

1993 2013 2033 2053


500 r---__

a: 300
UJ 200
> 100

o 10 20 30 40 50 60 70 80

Figure 9. The depletion of recoverable proved reserves of coal,

with an initial annual production of 800 million tons and
assuming a 5% annual increase in coal production and no
additions to the reserves.

the coal industry. These include physical restraints, logistic

transportation problems, environmental concerns, esthetic consider-
ations, and political limitations. In addition, interfuel substitu-
tion of coal for oil and natural gas can be costly and time consum-
ing. Much of the U.S. coal currently mined is in eastern deep mines,
and production costs of coal from these are about double those for
coal from surface strip mining. Both types of mines require devel-
opment times of a number of years prior to production start-up. The
produced coal is typically crushed, washed, and transported from the
mine via railroad. There is now a movement toward shipping the coal
by pipeline, where the coal is pumped as a coal/liquid slurry. The
liquid can be water, methanol [33], liquid C02 [34], or hydrocarbon
solvent [35,36]. Although coal/oil mixtures have been tested commer-
cially in electric power plants [37], this technique of coal trans-
portation is still principally in the research and development stage.

Coals contain appreciable concentrations of heteroatoms, and

some of these can cause deleterious environmental effects. Legis-
lated standards for sulfur emissions have been in effect for 15 years,
and these curtailed the direct burning of most coals by industries
without either coal clean-up or exit gas and water scrubbing. Pend-
ing legislation would reduce the allowable sulfur content of coal by

Table 8. A Scheme of Consumption of the U.S. Coal Reserves

(billions of tons)

Annual Final Year Net Consumed

Time Period Production Annual with Constant
Growth Rate Production 5% Growth

1983-2008 5% 2.71 40.89 40.89

2009-2018 4% 4.01 74.71 76.67
2019-2028 3% 6.01 123.86 134.94
2029-2038 2% 7.32 190.94 229.84
2039-2048 1% 8.09 268.31 384.42
2049-2058 0% 8.09 349.19 636.22

about half, and would, in effect, prohibit the direct burning without
clean-up, e.g. in coal-fired electric power generating plants, of
all bulk coals in the U.S. It is often stated that western U.S.
coals are very low sulfur coals. However, most western coals have
apparently not been intensively explored nor analyzed. The western
coals that have been utilized and fully characterized are the highest
grade available, while western coals containing up to 10% sulfur are
known. Therefore, as time and production proceed, the quality of
mined coal can be expected to decline. Therefore, it follows from
the preceeding discussion that the actual future coal utilization
will likely be less than indicated in Table 8, and the growth rate
in coal production might be on the order of 3% during the next 45
years, which will lengthen the life-time of the U.S. reserves.


Attempts in forecasting productivity, advances in stock prices,

production of mineral resources, consumption of energy, etc. can be
frustrating at best. However, studying past forecasts and measuring
actual performance against them can be instructive in determining
progress toward a goal. In regard to the four specific areas listed
above, the literature is full of forecasts and trend analyses. Energy
production and consumption is of interest here, and a few U.S. gov-
ernment projections will be used to gain insight into trends.

It is evident that estimates made by the U.S. Bureau of Mines

in 1975 (Table 7) were entirely inflated in regard to the usage of
crude oil and the total quantity of energy consumed because the price

increases of 1979-1980, as well as the conservation by the public

(Table 6) and the increase in efficiency by industry (Table 5), could
not be foreseen. In November 1980, with knowledge of these occur-
rences, the U.S. Department of Energy made projections of energy
production and consumption for the years 1985 and 1990 [38]. The
projections were intended as baseline projections with as few assump-
tions made as possible. Therefore, the assumption that was made was
that government policies were to remain unchanged. The resultant
projection for 1985 is shown in Figure 10.

The total energy of 41.1 million barrels per day of oil equiv-
alent (MMBDOE) that is consumed corresponds to 87.0 Quads. This
energy consumption is broken down into the four principal U.S. sec-
tors in Table 9 and is compared with the energy utilization of 1980.
These numbers can be compared with those derived from the 1990 pro-
jections contained in Figure 11. The first observation to be made is
that the total energy projected to be consumed in 1985 is much too
high. Although the energy usage was 75.9 Quads in 1980, the reces-
sion resulted in a decline to only about 71.8 Quads being consumed
in 1982 (Table 7). This will increase by about 1.5% in 1983. Even
with 5% increases in 1984 and 1985, the total annual energy consump-
tion would only reach 80.34 Quads, which would be 1.4 Quads above
the 1979 level (Table 5).

It appears from Table 9 that every sector, other than transpor-

tation, will experience a large growth between 1980 and 1985. While
this is true, it should be pointed out that the energy consumption
by the industrial sector will be less in 1985 than it was in 1973
(compare Tables 6 and 9). This is an indication of the enhanced ef-
ficiency of U.S. industry, as well as its greater reliance upon elec-
tricity as an energy source. Energy consumption by the transporta-
tion sector will show little growth because it is a mature sector in
the U.S. and because of the mandated progress in enhancing efficien-
cies. In addition, increased prices encourage conservation, which
is easier to accomplish in this sector than in the other three

It appears that the projected energy consumption for electric

power generation in 1985, as well as 1990, has been over-estimated.
The data in Figure 10 indicate that if electricity usage is decreased
by 0.5 MMBDOE (from the 4.1 out), a net savings of 3.6 Quads of total
consumed energy will be effected. This "savings" might occur because
of the recent slack demand for electrical power.

From this review of energy utilization projections, it can be

seen that the actual energy consumed in the U.S. has been consistent-
ly less during the last decade than had been forecast. The implica-
tions of this are (i) the U.S. can be more energy self-sufficient
and (ii) the energy resources of the U.S. will last for a longer
period of time.


!Lf:crl'll(;lrr TR.I\N5MISStOH
G£N£RAfION "0$$1$
1) t '''1






'IilANSPOAT" no'f

Figure 10. Department of Energy projected U.S. energy sources and

uses in 1985, given in millions of barrels per day of
oil equivalent (MMBDOE).


The chemical industry is currently a petrochemical industry,

although it was previously noted that certain chemicals, e.g. meth-
anol, formaldehyde, ammonia, and urea [15] are obtained from natural
gas. Petroleum is the least abundant fossil fuel in the U.S., and
it is expected that it will be the first one to be depleted. At the
beginning of this century, coal was the source of many chemicals and
materials, but it was replaced by readily available crude oil. Today,
all of the large number of organic commodity chemicals are potential-
ly accessible via synthesis gas produced from coal by gasification.
It should be noted that renewable biomass could be analogously uti-

Following the oil embargo of 1973, research efforts intensified

in the area of producing chemicals by direct liquefaction of coal and
by the indirect route of gasifying coal to synthesis gas. Initially,
it was expected that a doubling in the price of crude oil would make
the coal liquefaction/gasification routes to chemicals competitive
with petroleum refining. However, from 1973 to 1976 the price of
crude oil quadrupled (Figure 3). And then from 1976 to 1980, the

Table 9. Energy Consumpti.on of 1980 compared with Projected Usage

in 1985 and in 1990 (in Quads)

1980 1985 a 1990a

Electricity Generation 24.4 29.4 34.7

Transportation 19.7 19.9 20.1
Industrial & miscellaneous 20.9 23.1 24.1
Residential & commercial 11.0 14.6 13.3
Total 75.9 87.0 92.2

aCalculated from reference [38]

price tripled again, but the competitive point was not achieved.
This phenomenon is a case of the "receding break-even point" [39].

While the price of crude oil was increasing, the prices of other
resources were also increasing. Some of these, e.g. all of the energy

(l(CTItiClf¥ lOSS E5
". IN
.. I Ql.jl



Figure 11. Department of Energy projected U.S. energy sources and

uses in 1990, given in millions of barrels per day of
oil equivalent (MMBDOE)
resources, were increasing faster than the rate of inflation because
of high demand (and later because of deregulation). The basic chem-
istry and process engineering for converting coal into chemicals had
not progressed far enough for immediate plant start-up to proceed.
In addition, the conversion processes for petroleum and for coal are
much different, so plants could not be simply retrofitted. There-
fore, new plants would have to be built for coal liquefaction, coal
gasification, and for transforming synthesis gas into chemicals. Of
course, the plants would be constructed of components produced from
raw materials costing inflated dollars, and the plants would utilize
feedstocks costing inflated dollars. This is further magnified by
the fact that processes that have been developed for transforming
coal into commodity chemicals are very capital intensive and must
generally be planned for a massive scale. Therefore, as research,
development, construction, and plant start-up proceed, the break-even
point continuously recedes.

However, once the plants are established, the costs of raw mate-
rials for coal gasification and for synthesis gas conversion are
projected to be lower than comparable chemical syntheses from petro-
leum [40]. A projection of the relative costs [40] for some chemical
feedstocks in 1990, based on the price per 1b of carbon contained in
the feedstock, is given in Figure 12. On this basis, 1:1 synthesis
gas and propylene will be comparable in price, as will 2:1 synthesis
gas and ethylene. However, the synthesis gas mixtures contain oxygen,
and this results in the real cost of 1:1 synthesis gas to be only
40% of the cost based on carbon content alone. Therefore, it is ex-
pected that in 1990 the relative costs of the chemical intermediate
feedstocks will increase in the order 1:1 synthesis gas < CO < pro-
pylene < 2:1 synthesis gas < ethylene < methanol.

The utilization of these feedstocks will depend on the value of

the end-products and efficiencies of reaction. Ideally, all of the
reactants should be found in the product at the end of a process,
and water or carbon dioxide should not be formed as side-products.
Table 10, partially derived from [40], compiles the reaction stoichi-
ometries for some of the products that can be formed from synthesis
gas. High efficiency is noted by a usage ratio of 1.0. Based on
this table and the relative feedstock costs indicated above, the eco-
nomic potential for the chemicals listed in Table 10 increase in the
order octane < ethylene < ethanol < acetaldehyde < acetic anhydride <
methyl acetate < acetic acid < ethylene glycol. The actual commer-
cialization of processes producing these products from synthesis gas
will depend on factors such as development of catalyst systems, suc-
cessful reactor engineering, solving corrosion and materials prob-
lems, and governmental incentives.

Although octane is the lowest on the list of economic potenti-

ality, it was the first process commercialized and SASOL had plant
start-ups in 1955, 1981, and 1982 to produce gasoline from coal via

Para form

Formaldehyde (aq)
Ethylene 2:1 Synthesis Gas
Propylene 1: 1 Synthesis Gas
Carbon Monoxide
Natural Gas
Crude Oil
Vacuum Residuum Coal

Figure 12. Projected relative 1990 costs (price/lb carbon) of raw

materials and basic chemicals, assuming "normal" price
increases for fossil fuels.

synthesis gas. In addition, a methanol (via synthesis gas from nat-

ural gas) to gasoline plant, using Mobil technology and the ZSM-5
catalyst, is slated for start-up in 1985 in New Zealand with a
12,500 barrel/day production volume. Of the other chemical products
given in Table 10, acetic acid is currently being synthesized by the
Monsanto methanol carbonylation process [41], and methyl acetate and
acetic anhydride will be produced from coal via synthesis gas and
methanol in late 1983 by Tennessee-Eastman [42].

Ethylene Glycol

Intensive research has been carried out in attempts to produce

ethylene glycol in high yield with high selectivity from synthesis
gas. Ethylene glycol is currently typically produced from ethylene
derived from petroleum by two-step processes involving oxidation to
ethylene oxide, which is then hydrated. The over-all reaction is
often considered to be


However, the first step of the process involves a series of two re-
actions, (2) and (3), that limit the maximum yield of ethylene oxide,
and subsequently ethylene glycol, to 86% [43].

6CH2CH2 + 602 -+ 6C2H40 + 60(ads) (Z)

CH2CHz + 60(ads) -+ 2C02 + 2H20 (3)

New processes that might be utilizable [44-50] include (i)

direct synthesis from synthesis gas (equation 4), (ii) reaction of
an acetal (formed from methanol by acid catalysis) with CO to yield

Table 10. Reaction Stoichiometries for the Formation of Selected Chemicals from Synthesis Gas
and Methanol

Product Reaction H2/ CO Usagea

Methanol 2H2 + CO ~ CH30H 2.0 1.0

Ethanol 4H2 + 2CO ~ CH3CH20H + H20 2.0 1.4
CH30H + 2H2 + CO ~ CH3CH20H + H20 2.0 1.4
Acetaldehyde 3H2 + 2CO ~ CH3CHO + H20 1.5 1.4
CH30H + H2 + CO ~ CH3CHO + H20 1.0 1.4
Acetic acid 2H2 + 2CO ~ CH3C(O)OH 1.0 1.0
CH30H + CO ~ CH3C(O)OH 0 1.0
Ethylene 4H2 + 2CO ~ CH2CH2 + 2H20 2.0 2.3
Ethylene glycol 3H2 + 2CO ~ HOCH2CH20H 1.5 1.0
Methyl acetate CH30H + CH3C(O)OH ~ CH3C(O)OCH3 + H20 0 1.2
Acetic anhydride CH3C(O)OCH3 + CO ~ CH3C(O)OC(O)CH3 0 1.0
Octane l7H2 + 8CO ~ CH3(CH2)6CH3 + 8H20 2.1 2.3
aReactant usage wt of feed/wt of desired product

dimethy10xa1ate, which is then reduced, (iii) reaction of methanol

with synthesis gas (equation 5), (iv) direct reaction of methanol
with formaldehyde (equation 6), and (v) carbony1ation of formaldehyde
to yield glycolic acid, which is then esterified to methyl glycolate
followed by reduction (equation 7).

3H2 + 2CO + HOCH2CH20H (4)

CH30H + CO + H2 + HOCH2CH20H (5)

CH30H + HCHO + HOCH2CH20H (6)

HCHO + co + H20 + HOCH2COOH

2H2 (7)

The commercially successful process will combine moderate capital

requirements and minimum feedstock clean-up with a stable catalyst
system producing high activity and selectivity.


In this discussion of the formation of chemicals from synthesis

gas, it can be seen that methanol is often a pivotal central chemical
intermediate. Although methanol is formed from a H2-rich synthesis
gas, it has a desired reactant ratio of 1.0 (Table 10). Methanol is
perhaps the most selectively formed commodity chemical produced, and
it is synthesized by a well-established process that utilizes the
rather mild reaction conditions of 50-100 atm pressure and 250°C
[51-53]. The catalyst for this synthesis consists of an intimate
mixture of Cu/ZnO [54,55] that is prepared by coprecipitation and re-
duction, and which can be supported on alumina or chromia. This
catalyst can be deduced to be one of the most active catalysts known

The 1983 u.s. production of methanol will be at least 1.3 bil-

lion gallon [57], and it has been projected that growth rates for
methanol production will exceed those of other major chemical inter-
mediates over the next decade. The size of the plants that are now
coming on-stream, e.g. the ARCO Chemical methanol plant at Channel-
view, TX [58] and the Celanese Canada plant near Edmonton [59], are
such that the output capacity is on the order of 200 million ga1/yr
(-2300 short ton/day). It should be noted that this is approximately
the maximum size of conventional methanol synthesis reactors [60].
Improved technology is needed so that 5000 or 10,000 short ton/day
reactors can be utilized. The use of one 5000 ton/day reactor, in-
stead of two conventional 2500 ton/d~y reactors, in a methanol syn-
thesis plant would yield over a $30 million savings in capital in-
vestment [60]. Note that the plants converting coal to synthetic
fuels, referred to in the Coal section of this paper, would produce
about 6800 short ton/day of oil-like material or approximately 6900
short ton/day of methanol. Some recent progress has been made in
the technology of methanol production, e.g. Prism separators can
recover hydrogen and carbon dioxide from the purge stream, which in-
creases the production capacity of existing facilities by 2.6% [61].

Some of the products that are or can be synthesized from meth-

anol are listed in Figure 13. About 30% of the U.S.-produced meth-
anol is oxidized to formaldehyde, while another 10% is converted to
acetic acid [57]. In comparison, as much as 50% of the world meth-
anol production is directed to the synthesis of formaldehyde [62].
In addition to these uses, methanol can also be utilized for the
(a) synthesis of lower olefins or aromatics,
(b) an indirect fuel additive by conversion to methyl-tertiary-
butyl ether (MTBE),
(c) directly as an automotive fuel additive,
(d) formation of methylamines,
(e) formation of halo-methanes,
(f) a fuel for electric power peak shaving, and
(g) a feedstock for the growth of single cell protein.
Many of the catalytic conversion processes, included in the 12 uses
listed above and in Figure 12, are described elsewhere in this volume.

Specialty Chemicals

Beginning in mid-1982, it became evident that many U.S. chemical

companies were diversifying in the sense of changing their emphasis
from bulk commodity chemicals to specialty chemicals. During the
prior decade, it had become clear that the specialty chemical market
had weathered the recessions and oil shortages better than the com-
modity chemical market had. This is reflected by a high steady
growth that is projected to be enhanced even further during the next
decade [63]. The smaller chemical firms have emerged from the 1970s
in a strong position because of the high margins yielded by the spe-
cialty chemicals, and these firms are now facing competition from
the major chemical companies. Some of the markets being strongly
contested include those utilizing water-treatment polymers, fluoro-
carbons, metal coatings, and electronics chemicals, and these could
grow at annual rates of up to 15%. Specialty chemical producers will
be seeking the most economical feedstocks availabl·e as competition
increases further, and many of these fall back to the basic CI-C3
building blocks.

An interesting digression is that the Soviet economist Nikolai

Kondratieff has pointed out that capitalist economies have followed
a regular 50-year cycle of development in which an exceptional clus-
ter of new inventions is followed by a period of economic upswing in
which the inventions are utilized [64]. This has been verified by

lwa.. r



-1 Cu/ZnO





Figure 13. Valuable chemicals can be produced from natural gas via
synthesis gas and methanol.

Gerhard Mensch, who estimates that approximately half of the basic

inventions of the current cycle have already reached the commercial
feasibility stage [64]. The decade of maximum innovation should
start in 1984, with 1989 being the year of peak innovation. This
period of maximum innovation will be accompanied with high energy and
feedstock costs, and in order to support the necessary research and
development, the top priority of U.S. industries will be to improve
profitability. This can be accomplished by (i) turning to and empha-
sizing specialty chemicals and (ii) elimination of much of the cur-
rent overcapacity with many commodity chemicals. Belief in this
trend of cycling shall further encourage us to optimistically pursue
the scientific and technological advances necessary to provide for
a diverse and continuing supply of chemicals and energy, which is the
basis for our high standard-of-1iving in the United States.

The chemical industry consumes approximately 10% of the U.s.
energy, and most of it is derived from petroleum. The fluctuations
in the price and availability of petroleum during the last decade has
provided the initiative for seeking alternative feedstocks and alter-
native sources of power. With the recent demands made for interfue1
substitution and inter-feedstock switching, a renewed focus on syn-
thesis gas and alcohols has appeared because of their capabilities
to satisfy both demands. Because of the versatility of these feed-
stocks, it is expected that the growth in production of these will
continue, as will the research and development in methods of utili-


1. A. B. Meine1 and M. P. Meinel, in "Ethics and Energy," Edison


Electric Institute, Washington, D.C., 1 (1979).

2. P. B. Weisz, CHEMTECH, 10(1):7 (1980).
3. J. B. Kirkwood, in "Energy for Survival," ed. by H. Messel,
Pergamon Press, Rushcutters Bay, Australia, 119 (1979).
4. R. W. Barnes, CHEMTECH, 8(1):30 (1978).
5. ''World Energy Outlook," Organization for Economic Cooperation
and Development/International Energy Agency, Paris, 354
6. "Energy Balances of OECD Countries," lEA; as reported in Refer-
ence 5, pp 331-333.
7. G. C. Gambs, ACS Symp. Ser., 110:135 (1979).
8. Coal Age, 88(7):40 (1983).
9. "The World Almanac and Book of Facts," Newspaper Enterprise
Assoc., Inc., New York, pp 140-146 (1983).
10. Associated Press release, "The Morning Call," Allentown, PA,
A15 (Aug. 4, 1983).
11. "International Monetary Fund and European Economic Community,"
as reported in Reference 5, pp 80-82.
12. Oil Gas J., 78(52):78 (1980).
13. "Petrole Informations," as reported in Reference 5, pp 208-212.
14. "Historical Review of Domestic Oil and Gas Exploratory Activi-
ty," DOE/EIA-0196 (October 1979).
15. Chem. Eng. News, 61(14):16 (April 4, 1983).
16. Oil Gas J., 81(23):25 (1983).
17. Information provided by the UGI Corp. (Pennsylvania), July 1983.
18. ''World Energy Conference" (1978); as reported in Reference 5,
p. 287.
19. ''World Coal Study, Coal-Bridge to the Future," Ballinger, Cam-
bridge, MA (1980).
20. J. B. Kirkwood, in "Energy for Survival," ed. by H. Messel,
Pergamon Press, Rushcutters Bay, Australia, 139 (1979).
21. W. Glasgall, Associated Press, "Call-Chronicle," Allentown, PA,
A14 (Aug. 31, 1980).
22. "Facts About Oil," American Petrolewn Institute~ Washington,
D.C., 41 (1981).
23. "The Power of Coal," National Coal Association, Washington,
D.C., 10 (1981).
24. B. Dvorchak, Associated Press, "Call-Chronicle," Allentown, PA,
B12 (May 31, 1983).
25. "Statistical Abstracts of the United States," 103rd Ed., U.S.
Department of Commerce, pp 572-573 (1982).
26. "World Energy Outlook," Organization for Economic Cooperation
and Development/International Energy Agency, Paris, 300-301
27. P. L. Layman, Chem. Eng. News., 60(48):9 (November 29, 1982).
28. "Report of the High Level Group for Energy Technology Commer-
cialization," Organization for Economic Cooperation & Devel-
opment/International Energy Agency, Paris (1981).
29. "United States Energy Through the Year 2000," U.S. Bureau of
Mines (December 1975).

30. oil Gas J., 81(5):71 (1983).

31. J. F. Wunderlin, Ind. Heating, L(8):34 (1983).
32. D. L. Rainey, O. C. Sheese, and M. D. Bullen, Ind. Heating,
L(5):6 (1983).
33. Coal Age, 86(11):33 (1981).
34. Chem. Eng. News, 59(42):34 (Oct. 12, 1981).
35. J. G. D. Schulz andJ. A. Cobler, U.S. Patent 4,272,253 (1981);
assigned to Gulf Research and Development Co.
36. Ind. Heating, IL(1):12 (1982).
37. Chem. Eng. News, 60(32):16 (Aug. 9, 1982).
38. "Notice of Public Hearings & Staff Working Paper," DOE/PE-0022
(March 1981).
39. J. Haggin, Chem. Eng. News, 59(30):43 (July 27, 1981).
40. A. Aqui16, J. S. Alder, D. N. Freeman, and R. J. H. Voorhoeve,
Hydrocarbon Process., 62(3):57 (1983).
41. J. F. Roth, J. H. Craddock, A. Hershman, and F. E. Paulik,
CHEMTECH, 1:600 (1971).
42. J. L. Ehrler and B. Juran, Hydrocarbon Process., 61(2):109
(1982) •
43. P. A. Kilty, N. C. Ro1, and W. M. H. Sacht1er, Proc. 5th Intern.
Congr. Cata1., Vol. 2, ed. by J. W. Hightower, North-Holland
Pub1. Co., Amsterdam, 929 (1973); and H. T. Spath, Ibid.,
945 0.973). --
44. S. Suzuki, U.S. Patent 3,911,003 (1975); assigned to Chevron
Research Co.
45. L. A. Cosby, R. A. Fiato, and J. L. Vidal, U.S. Patent 4,115,433
(1978); assigned to Union Carbide Corp.
46. L. R. Zehner and R. W. Lenton, U.S. Patent 4,112,245 (1978);
assigned to Atlantic Richfield Co.
47. R. C. Williamson and T. P. Kobylinski, U.S. Patent 4,170,605
(1979); assigned to Gulf Research and Development Co.
48. J. F. Knifton, J. Am. Chern. Soc., 103:3959 (1981).
49. M. B. Sherwin, Hydrocarbon Process., 60(3):79 (1981).
50. S. Suzuki, J. B. Wilkes, R. G. Wall, and S. J. Lapporte, in
this volume.
51. R. G. Herman, K. Klier, G. W. Simmons, B. P. Finn, J. B. Bu1ko,
and T. P. Kobylinski, J. Cata1., 56:407 (1979).
52. R. G. Herman, G. W. Simmons, and K. Klier, Proc. 7th Intern.
Congr. Cata1., ed. by T. Seiyama and K. Tanabe, Elsevier,
Amsterdam, 475 (1981).
53. K. Klier, V. Chatikavanij, R. G. Herman, and G. W. Simmons,
J. Catal., 74:343 (1982).
54. S. Mehta, G. W. Simmons, K. Klier, and R. G. Herman, J. Cata1.,
57:339 (1979).
55. J. B. Bu1ko, R. G. Herman, K. Klier, and G. W. Simmons, J. Phys.
Chem., 83:3118 (1979).
56. K. Klier, Adv. Cata1., 31:243 (1982).
57. Chem. Eng. News, 61(4):27 (Jan. 24, 1983).
58. Chem. Eng. News, 60(33):9 (Aug. 16, 1982).
59. Chem. Eng. News, 61(4):22 (Jan. 24, 1983).

60. M. E. Frank, CHEMTECH, 12(6):358 (1982).

61. B. M. Burmaster and D. C. Carter, Oil Gas J., 81(13):90 (1983).
62. R. Pearce and M. V. Twigg, in "Catalysis and Chemical Proc-
esses," ed. by R. Pearce and W. R. Patterson, Wiley & Sons,
Scotland, 114 (1981).
63. N. Furstinger, Ind. Chern. News, 4(6):24 (1983).
64. M. Hyde, Ind. Chern. News, 3(12):1 (1982).

Michael S. Graboski

Colorado School of Mines

Chemical and Petroleum-Refining Engineering Department
Golden. Colorado 80401


Synthesis gas. composed principally of carbon monoxide and hydro-

gen can be used as the major building block in the production of
chemicals and fuels. Natural gas, petroleum liquids, biomass and
coal may all be readily reformed or partially oxidized to produce
synthesis gas suitable for further processing. Most synthesis gas
produced today comes from natural gas by catalytic reforming. but
commercial production of synthesis gas by partial oxidation of heavy-
liquids is also widely practiced. Only a minor amount of synthesis
gas is produced today by solid fuel gasification.

The term synthesis gas is applied to a gas mixture containing

mainly CO, HZ' and COZ' with minor levels of other components, such
as methane. In synthesis gas, an important ratio from the end use
point of view is the HZ/(co+coZ) ratio of the gas. Using the variety
of feedstocks available and state of the art technology, it is
possible to prepare any synthesis gas composition desired. For
example, in methanol manufacture with natural gas feedstocks, the
synthesis gas is rfch in hydrogen. Usually the synthesis gas compo-
sition is altered by addition of C02 from an external source to
maximize hydrogen utilization. On the other hand. if synthesis gas
is produced from heavy oil or coal. the gas is rich in CO; here water
gas shifting is used to convert externally added steam to hydrogen
at the expense of CO.


Steam reforming of natural gas to produce hydrogen and synthesis


gas for methanol and ammonia synthesis has been practiced commercial-
ly since 1930. It is the most widely used method of hydrogen manu-
facture in the United States. The major reformer feedstocks are
natural gas, propane, LPG, butane, naphtha and refinery off-gas [1].

The primary reactions carried out in methane steam reforming are

as follows:

+49 kcal/mole (1)

co + H20 -+ C02 + H2 (2)

Under reforming conditions using modern catalysts the reactions can

be brought close to thermodynamic equilibrium while suppressing coke
deposition. Heavier feedstocks can be reformed in the same manner
yielding different quantities of CO andH2 based on their composition.
Theoretically, 50% of the (CO + H2) obtained from natural gas comes
from water decomposition, while about 35% comes from water decomposi-
tion in naphtha reforming. Kermode [2] gives a good discussion of
methane steam reforming, while Casper [3] provides a typical energy
balance. Marschner [4] provides a good discussion of process varia-
tions in naphtha reforming, and Kermode [2] gives naphtha energy
balance data.

A typical flow sheet for a hydrogen steam-reforming plant is

shown in Figure 1. The main components include desulfurization,
reforming and heat recovery. The sulfur recovery method employed
depends on the feedstock [4-6]. For natural gas feeds, metal impreg-
nated activated carbon or a combination of hot zinc oxide (>300°C)
and activated carbon can lower the sulfur content to 0.2 PPM. This
low level of sulfur insures long catalyst life. Desulfurization
of other feedstocks is more difficult. For hydrocarbon stream (>C2)
containing organic sulfur and olefins, generally a cobalt molybdenum
hydrodesulfurization catalyst is employed to hydrogenate sulfur to
H2S and to saturate olefins, where the hydrogen is supplied by recy-
cling some of the reformer effluent. With naphtha, a pretreatment
with 98% H2S04' followed by caustic and water washes, is used to
remove soluble organic sulfur and olefins. The naphtha is then
vaporized and treated over CoMo and ZnO.

The reformer consists of a furnace and tube bank. The heat

required for the endothermic reforming reaction is supplied by radiant
heat transfer across the tube wall. Furnaces may be gas or oil fired.
Wang [7] discusses conversion of gas fired units to oil. For good
heat transfer, tubes are designed with inside diameters between 2.75
inches and 6 inches II] and lengths between 32 and 44 feet. Typical
pressurized units have 4 inch ID tubes with 3/4 inch wall thickness
[2]. Some of the major furnace types and manufacturers are: Kellogg
(down-fired), Foster-Wheeler (terraced, side-fired), Selas (side-fired)
and Chemico (up-fired). A good discussion of modern reforming hard-




Fig. 1. Natural Gas Reforming for the Production of Synthesis Gas.

ware is given by Marsch [8]. Feedstock is preheated to 1000 to 1100°F

(about S400 to 600°C), using furnace waste heat and feedproduct ex-
change. The reformer exit temperature usually does not exceed l6S0°F.
The flue gas in the fired section of the furnace is usually 1800 to
1900°F and exits the furnace stack at 400 to SOO°F. The reforming
pressure ranges from atmospheric to 600 psig. The most common refor-
ming pressure is in the 300 psig range. Tubes and ancilliary high
temperature piping are manufactured from high Nickel-Chrome alloys,
and HK-40 (20 Ni-2SCr) is generally used. Alternately, Inconel 617
(76Ni, l6Cr, Fe) Incolloy 800 and 807 (32Ni, l6Cr + Fe), and Super-
therm (3SNi-2SCr) may be used [9]. Tube life of 100,000 hr is possi-
ble, but in pressurized service the life is decreased by 50% for each
SOoF over the design temperature in continuous service [S].

The reforming reactions closely approach equilibrium. The meth-

ane-steam reaction is usually within 20°C of the reformer exit tem-
perature while the shift reaction is equilibrated at the reformer
exit temperature. Table 1 gives a typical composition of reformed
methane. Based on equilibrium, the synthesis gas composition for a
given feedstock, steam and CO 2 input can be readily calculated.
Kermode [2] presents some equilibrium predictions and provides
references for additional sources of data. Morse [10] gives a simpli-
fied design method based on equilibrium considerations. According to
equilibrium, as pressure is increased, the reforming reaction is
suppressed. Therefore, all things being equal, a greater degree of
methane leakage will occur from pressurized reformers. Because of
strength of material and cost considerations, pressurized reformers
tend to run at lower exit temperatures than atmospheric reformers.
In hydrogen plants operating at 300 psig, for example, the exit gas
temperature is maintained at around lS00°F (8lS0C), while at atmos-

Table 1. Composition of Synthesis Gas,

% by Volume

Component -95% Conversion Heavy Oil

of Methane Oxidation

CO 15.5 47.0
COZ 8.1 4.3
HZ 75.7 46.0
NZ+A O.Z 1.4
CH4 0.5 0.3
HZS o 1.0

pheric pressure, temperatures as high as 1650°F" (900°C-) are main-

tained. The economics of synthesis gas production suggest that re-
forming should be done at the use pressure if possible. Carbon·
deposition is of concern in reformers and must be suppressed. White
[11] provides a discussion on the thermodynamics of carbon deposition.
Prediction of the carbon deposition boundary by equilibrium calcula-
tions generally gives optimistically low steam or steam + COZ/feed-
stock ratios. Practically, molar steam to carbon ratios of 3.5 to
4.5/1 are used. Carbon dioxide can be substituted for some steam
if desired. Carbon deposition will generally deactivate the catalyst
causing local hot spots in the tube wall leading to tube failure.
Olefins are to be particularly avoided in feeds due to their coking
tendency. Increasing the steam and/or COZ rate has the beneficial
effect of further decreasing the methane leakage. Grover [lZ] pre-
sents a useful kinetic model for reformer design, while Dorga [13]
presents useful case study information on existing units. Hydrogen
and methanol synthesis generators usually produce gases with a methane
content of 0.5 to 3%.

Catalysts generally consist of nickel (6 to ZO weight %) on a

ceramic carrier such as alumina, calcium-aluminate or magnesium-
aluminate; the most. common size and shape being l/Z in. to 3/4 in.
Rashig rings [1]. A typical design space velocity is 8000 V/V hr of
wet feed. Catalyst life is dependent on sulfur and coking. However,
under normal operating conditions, at least several years of service
are obtained per charge.

Uchida and Kyogoku [14] report that Z3.5 lb of naphtha are

required per 1000 SCF of 98% HZ at Z70 psig and l5Z6°F; where 56.4%
is reformed while 43.6% is burned to supply furnace heat. For natural
gas at the same conditions, 5Z5 SCF of high methane gas is required
with 5Z% being fed to the furnace. Overall, reformer efficiency

depends on heat recovery, with waste heat in the convection section

of the furnace and in the product gas being used to produce high
pressure steam for reaction and power and add feed preheat. Much of
the modern reforming developments center around energy optimization.

Synthesis gas for organic chemical production generally requires

H2/CO ratios between 1 and 3. Reformer effluents exhibit H2/CO ratios
of about 5 to 10, but C02 addition to the reformer feed will lower
this ratio. Carbon dioxide addition to the reformer is practiced in
low pressure operation. In high pressure cases, C02 is mixed down-
stream of the reformer. In methanol synthesis, the catalyst back-
shifts C02 [15] so that no further processing of the syntheis gas is
necessary. In other situations, it is conceivable that the appropriate
H2/CO ratio must be obtained prior to synthesis. Presently, catalytic
water gas shift (equation 2) is practiced commercially over iron-chrome

CO + H20 -+ CO 2 + H2 (2)

catalysts in low sulfur environments at temperatures between 600°F

and 900°F to produce hydrogen from CO. Backshifting is possible to
increase CO content at the expense of C02 and hydrogen, but this is
not practiced commercially since methanation is also favored under
these conditions. Reformer configurations involving C02 recovery
and recycle (and possibly addition) can produce adjusted H2 /CO ratios,
but uneconomically.


Gasification is the term commonly applied to the partial oxidation

of fuels. Generally, the following set of reactions are carried out:

Fuel + 02 -+ CO + H2 + Heat (3)

Heat + Fuel + Steam -+ CO + HZ (4)

CO + HZO -+ CO 2 + HZ (Z)

Depending on the fuel being processed, the actual full set of reac-
tions occurring is quite complicated since thermal pyrolysis is an
important reaction step. In gasification systems, the overall objec-
tive is to maximize the chemical energy of (CO + HZ) and minimize
methane in the final product gas. To accomplish this, the fuel is
processed at high temperature in an oxygen deficient environment,
with heat released by the oxidation used to raise the reaction mix-
ture temperature and supply energy for the endothermic reforming

Gasification of liquid and solid fuels generally occurs in short

residence times (1 to 10 sec), high temperatures (>1200°C at reactor

exit with higher temperature in the oxidizing zone) and pressures

ranging from 1 atm to 100 atm. Gasification has a number of advan-
tages for synthesis manufacture relative to steam reforming techno log)
These are the following:

1) The process is non-catalytic. Therefore, fuels containing

sulfur and other catalyst poisons can be readily converted to
CO and H2 .

2) Reaction conditions are usually severe enough so that there

is almost no methane leakage through the system.

3) The reaction chamber walls are refractory lined or water

cooled, which simplifies metallurgical requirements, particularly
for pressurized operation.

4) The equipment can be small.

The principal disadvantages are:

1) For solid fuels, feeding is difficult, especially to

pressurized systems.

2) High purity oxygen is required so that the inert diluent

level in the synthesis gas is small.

3) Heat recovery from the product gas is expensive.

4) The overall capital cost for producing synthesis gas of

desired purity is large.


The gasification of oil (naphtha, crude oil and residuum) by the

Shell and Taxaco processes has been practiced commercially for more
than 25 years. The basic differences between Shell and Texaco tech-
nologies are in burner design and downstream gas cooling and cleaning
equipment. Madsak [16] gives a brief history of Shell units, dealing
particularly with feedstock trends, showing that in recently ordered
plants, the goal is to use the heaviest ends including asphalt,
heavy vacuum resid and vacuum vis breaker resid. He states that over
80 MM NM 3 /day of hydrogen capacity is available through Shell gasifi-
cation technology. Figure 2 shows a flow sheet for a typical oil
gasification facility producing synthesis gas. The gasifier consists
of a burner mounted in a cylindrical refractory lined chamber typi-
cally 30 feet long. Oil is fed in the liquid phase to the burner
and subsequently sprayed downward into the reaction chamber. Oxygen
and steam [17] which is added as a thermal moderator, are mixed and
fed to the chamber in a manner that promotes good mixing of the oil



Fig. 2. Partial Oxidation of Oil to Synthesis Gas

and steam/02 in the reaction chamber. The burner designs are proprie-
tary and feedstock dependent. Back radiation from the walls and flame
vaporize the oil and initiate combustion. Approximately 35 percent
of the stoichiometric oxygen and 0.4 lb steam/lb oil are fed to the
unit. In the flame region, pyrolysis of the oil and combustion take
place at very high temperatures (150QoC). Methane is the principle
hydrocarbon produced in pyrolysis, and some methane passes through
this zone intact. Homogeneous thermal steam cracking in the soaking
zone of the reactor allows methane to decompose. This reaction ceases
below about l300°C. About 3% of the oil by weight leaves as carbon
(soot) in the product gas. Pressure does not affect the reaction
kinetics or composition markedly. Kermode 12] presents data on 02
requirements for different fuels. The gas composition can be esti-
mated by assuming water-gas shift equilibrium at the exit temperature.
Gas composition can be varied by adjusting the 02/steam/fuel mix, and
cold gas efficiencies of 75% are typical.

Table 1 presents a typical composition from heavy oil gasification

by Texaco or Shell gasification. Relative to steam-reformed gas,
there is considerably more CO and less hydrogen. Therefore, downstream
catalytic water-gas shifting is necessary for most synthesis gas appli-
cations. In the Shell process, gas from the gasifier is cooled in
proprietary design waste heat boilers to recover high quality steam,
which is useful for process drivers, oxygen manufacture and process
steam. Soot is recovered in a scrubber and recycled to the gasifier
I17]. In the Texaco process, the gas is quenched with a hot water
spray in the base of the gasifier to wash out particulates and satur-
ate the gas with steam for water-gas shifting I18]. The soot is
recovered and added to the oil feed. Texaco also has a proven waste
heat boiler design.

Because of the sulfur in the gasifier effluent, water-gas shift-

ing to the final desired HZ/CO ratio is best accomplished using a
sulfided-cobalt-molybdate catalyst, such as that supplied by BASF.
Shifted gas is cooled using a variety of heat recovery methods and
cleaned completely of HZS and COZ to the desired level using liquid
scrubbing techniques. Christensen [19] discusses the merits of a
number of technologies based on pressure and COZ content. These
include Selexol, Rectisol, Benfield, Catacarb and DGA. There is a
good deal of commercial experience with all of these processes, and
guidelines for selection of the proper process are given. Kohl [ZO]
gives design data for a wide variety of gas cleaning techniques.


A number of processes are available for the production of synthe-

sis gas from coal. The Lurgi. Koppers-Totzek and Winkler methods have
all been proven commercially and are still being advertised on a world
wide basis. The processes are generally termed first generation and
are the forerunners of more efficient modern day technologies. In
this latter group, the Texaco gasification process, which is an adap-
tation of the oil gasification technology, and the KBW process, which
is based on Koppers-Totzek gasification, are most important. Many
new third generation technologies have been researched in recent years
including BGC slagging Lurgi, Hy-Gas, Bi-Gas, COZ-acceptor and West-
inghouse. None of these has been operated for an extended period of
time at large enough scale to be considered ready for commerical

Relative to oil, coal is more difficult and expensive to convert

to synthesis gas. The principle difficulties are due to (1) solid
form, (Z) high C/H ratiO, (3) high ash content, and (4) high S,N,Cl.
The solid form increases capital relative to oil because of grinding
costs and greater 0z requirements. Reliable feeding is difficult and
is currently accomplished by energy intensive lock hoppers and water
slurry systems, although a number of more cost effective feeder candi-
dates may replace these systems in the future [Zl]. High carbonI
hydrogen ratio and ash require the gasifier to operate at higher
temperatures or long residence times to obtain high carbon conversions
At high temperatures, the ash is slagged creating potential refractory
problems in systems not employing waterwalls. Ash and solidified slag
are abrasive and cause wear in letdown systems. Heat recovery is more
expensive in coal systems, where volatile ash components can condense
and collect on heat transfer surfaces and entrained ash can cause wear.
Most importantly, tar, waste water, sulfur, nitrogen and chlorine
recovery can represent a significant capital expenditure.

In terms of synthesis gas production, Texaco and KBW are the most
desirable modern day alternatives. While there are many years of
commercial operating experience with Lurgi at Sasol, South Africa
(Fischer-Tropsch to hydrocarbons), it should be stressed that Lurgi

is not an economical synthesis gas generator. With its updraft

(countercurrent) fixed (moving) bed 02 blown configuration, hydro-
carbon gases, tars, and liquids are distilled off of the coal at low
temperature and are preserved. The synthesis gas produced is rich
in hydrocarbons, with the result that significant gas treating is
required to bring the gas to the point where it can be utilized.
Because of the particularly high methane content of the gas, purge
streams in synthesis recycle loops tend to be large. The overall
effect is a highly capital intensive process. Lurgi has the advan-
tage of operating experience at the 300-450 psig level. Winkler and
Koppers-Totzek (K-T) were designed to produce a high quality synthe-
sis gas and employ a fluidized bed and entrained flow system,
respectively, to minimize hydrocarbon leakage. Winkler accomplishes
this by partial oxidation in the gasifier fluidized bed itself with
additional oxygen injection in the freeboa.rd of the gasifier to
further clean up hydrocarbons. The K-T process operates at very high
temperatures (about l700°C) in a combustion zone to destroy all hydro-
carbons produced. The major difficulty with both technologies is low
pressure operation (Winkler up to several atmospheres, K-T at atmos-
pheric pressure) necessitating high compression costs for the resul-
tant synthesis gas. Lurgi and Winkler are limited to low rank coals.
In the "think big" era of the 1970's when studies were based on 250
billion Btu/day synthesis gas plants, all three technologies were
looked on unfavorably because of the need for a large number of gasi-
fiers in a single plant. This is an unfair point of view because the
history of chemical plant development involves minimizing the capital
risk in the beginning of new corporate ventures by starting with small
scale commercial plants. The large energy factory concept has been
created by the government and will not be carried out in the private
sector without government assistance. Even so, Lurgi technology is
being installed in the U.S. Government backed 137 MMSCFD Great Plains
Coal Gasification project being constructed by CE Lummus to produce
synthetic natural gas (SNG). A methanol loop has been under consider-
ation in future planning.

The Texaco and KBW gasification systems have been deemed to be

commercially viable from an operating point of view. Texaco has,
for example, in planning or under construction, gasifiers at its
Montabello test site, Tennessee Eastman Project, TVA Muscle Shoals
Alabama Project, Coolwater Coal Gasification Project, and Ruhre Chemie
AG test site. The combined operating experiences of Texaco facilities
including information on feeding, refractory, heat recovery, pressu-
rized operation, waste-water analysis and operation on a wide range
of coal ranks makes Texaco the leader in coal-gasification technology
for synthesis gas production. However, KBW is considered to be
commercially viable based on no operating experience. Recently formed
by Koppers USA and Babcock and Wilcox, the KBW process is a modified
entrained pulverized coal boiler designed to operate fuel rich on
oxygen and coal. The technology is loosely related to Koppers K-T
process and B&W's gasification and boiler experience. It has been

classified as "commercial" with no operating experience by the

Synthetic Fuels Corporation (SFC) which has permitted KBW to
init.iate applications for loan and price guarantees on coal and
peat gasification processes.

The modern day third generation processes offer little hope of

commercial reality in the very near future. The BGC slagging Lurgi
offers promise as a commercial gasifier in the 1990's. By operating
a fixed bed with oxygen-lances at the bottom of the bed and no ther-
mal moderation, the gasification is complete and the ash is readily
tapped as a liquid. Temperatures are high enough to greatly reduce
the hydrocarbon evolution from the updraft gasifier. High methane
production is still a problem, although gasifier throughput is
increased dramatically over the conventional Lurgi. The Bi-Gas and
Hy-Gas process developments were misguided in that they were optimize,
for methane production. Thus, while capacity is potentially large,
the gas composition is unfavorable for chemical synthesis. The
Westinghouse process has been shown to be economically attractive
relative to other processes 122,23]. Westinghouse and SASOL have
recently entered into an agreement to install a Westinghouse unit
in Sasol, South Africa to gasify coal fines. Based on the believed
successful program, Westinghouse has recently replaced Texaco in a
partner's role in the New England Energy Park Project under applica-
tion to the SFC and plans to operate the gasification technology on
a large commercial scale to produce substitute natural gas and

According to Cornils 124], the Texaco process offers a number of

important advantages including high carbon conversion, low methane,
tar and phenols, low environmental impact, simple ash handling, high
throughputs, use of fine coal, and insensitivity to coal type. The
Texaco coal system is more capital intensive than the oil system.
Figure 3 shows a flow sheet of a typical system. Coal must be cleaned
and prepared to less than 100 mesh by either wet or dry grinding. In
the Texaco process, the coal is slurried with liquid water to produce
a slurry concentration of 50 to 70% solids. The slurry is preheated
and injected into the gasifier along with oxygen in a burner similar
to that used in the oil case. The liquid water is detrimental to the
process thermal efficiency, but it provides a reliable means of get-
ting the coal into the pressurized unit (up to 1500 psi). Reactions
similar to those in the oil gasification system occur with the addi-
tion of the freeing and slagging of the coal ash. This adds an
additional problem related to refractory life. In present designs,
the slag and gas are separated by a change in direction at the gasi-
fier base. The slag drops into liquid water, is quenched and removed
via a valving system. The hot gas passes through a waste heat system
and a final gas cleanup prior to going to water-gas shift and acid
gas removal. Table 2 shows some typical operating data for a large
Texaco gasifier. Relative to oil, the exit temperature is higher

....--........- . ; L - _ - ,

Drying Shift
& t---.j Gasification Converter t----,

Slag Ash




Distillation CompresSion

Steam Steam

Fig. 3. Route for Coal or Biomass to Methanol.

Table Z. Typical Performance Data for

Texaco Gasification on Coals [Z4]

Exit Temperature, °c 1300 to 1500

Pressure, atm 35
Gas Composition, Volume %
CO 55
HZ 33
COZ 11
CH4 .01
NZ 0.6
Carbon Conversion up to 99%
Gasification efficiency
coal gas up to 74%
thermal 9Z%

and results in lowering the cold gas efficiency. This is manifested

in a higher COZ content of the product gas. Even with the large wa-
ter/coal ratio, the gas is very rich in CO because of shift equili-
brium at the exit temperature. Corni1s [Z4] gives additional opera-

ting data. The Texaco gasifier has been operated at 150 tons per
day and this is equivalent to about 100 TPD of methanol which is
small by today's standards.


Biomass and peat may be readily gasified to produce synthesis

gas in coal systems or in process equipment designed specifically
to handle biomass. Graboski [25] presented a comparison of biomass
and coal as feedstocks and showed that biomass processes to make
synthesis gas are potentially less costly than coal processes. In
the biomass area, KBW [26] is currently involved in planning for a
1400 TPD (dry basis) peat to methanol plant using the First Colony
Farms peat reserves in Cresswell, North Carolina. The project is
under review by the SFC. Evergreen Energy Corp. [27] has announced
a joint project with Texaco to test the concept of gasifying wood
in the Texaco system. A commercial methanol plant based on wood
is being planned for New England. Methanol, Inc. (subsidiary of
International Harvester) is in the planning stage for a business
based on using packaged plants to produce methanol from wood using
a fixed bed down-draft gasification system developed at the Solar
Energy Research Institute [28]. The plant configurations are similar
to those for coal. The plant scale at which synthesis gas from
biomass is economical is considerably smaller (100 to 1000 TPD of
dry feedstock) than coal, which means that capital cost is less
than for coal plants. Biomass may therefore actually impact on the
market before the large coal systems that are currently in planning
and construction.


Synthesis gas, composed principally of carbon monoxide and hydro-

gen, can be used as the major building block in the production of
chemicals and fuels. Natural gas, biomass and coal may all be readi-
ly reformed or partially oxidized to produce synthesis gas suitable
for further processing. The purpose of this paper was to discuss
the state of the art of synthesis gas preparation from these feed-
stocks and to indicate current commercial ventures.

Most synthesis gas produced today comes from natural gas consti-
tuents. Alternately, synthesis gas may be produced from coal and
biomass. Suitable commercial-scale gasifiers, including Koppers-
Totzek, Texaco and Winkler gasification systems are applicable to
both coal and biomass processes. Because of the differences in
feedstock characteristics including reactivity, ash, sulfur, nitrogen
and chlorine, the process requirements for the two feedstock types
are considerably different. Typical process flow schemes for commer-
cial processes operating on coal and wood have been described.

A variety of commercial ventures are in the construction stage

in the United States. In the case of coal, these include the
Tennessee-Eastman chemical complex in Kingsport, TN, which will
begin to produce acetic anhydride from coal, via synthesis gas and
methanol, in late 1983. In the area of biomass utilization, the
KBW peat-to-methano1 plant is in the development stage and several
wood-to-methano1 plants are progressing through the planning stage.


l. P. Boyd, in Proc. Biomass-to-Methano1 Specialists Workshop,

SERI/CP-234-1590, Tamarron, Colorado, March 3-5, 1982.
2. R. 1. Kermode, "Hydrogen: Its Technology and Implications,"
Vol. I, CRC Press, Cleveland, Ohio, 1977.
3. M. S. Casper, "Hydrogen Manufacture by Electrolysis, Thermal
Decomposition, and Unusual Techniques," Noyes Data Corp.,
New Jersey, 1978.
4. F. Marschner and H. J. Renner, Hydrocarbon Proc., 61(4):176 (1982).
5. A. Longacre and H. Truby, Preprints, Div. Pet. Chem. (ACS),
6. R. N. Bery, Preprints, Div. Pet. Chem. (ACS), 16(2):C67(1971).
7. S. I. Wang, Hydrocarbon Proc., 58(4):193(1979).
8. H. D. Marsch and H. J. Herbort, Hydrocarbon Proc., 61(6):101(1982).
9. S. N. Narayan, J. Sci. Ind. Res., 32:664(1973).
10. P. L. Morse, Hydrocarbon Process, 52(1):113(1973).
11. G. A. White, T. R. Roszkowski, and D. W. Stanbridge, Hydrocarbon
Process, 54(7):130(1975).
12. S. S. Grover, Hydrocarbon Process., 49(4):109(1970).
13. L. S. Dorga, Chem. Age India, 20(1)(1969).
14. H. Uchida and H. Kyogoku, 68th AIChE Meeting, Los Angeles (1975).
15. M. J. Royal and N. M. Nimmo, Hydroca~:". 1.Rr:c:~ess. (48(3):147(1969).
16. H. J. Madsock, Hydrocarbon Process., 61(/:'·~'~982).
17. c. J. Kuhre and C. J. Shearer, Hydrocarbon Procetis., 50(12):113
(1971) •
18. W. G. Sch1inger and W. L. Slater, Preprints, Div. Pet. Chem.,
(ACS), 16(2):C45 (1971).
19. K. G. Christensen and W. J. Stupin, Hydrocarbon Process., 57(2):
125 (1978).
20. A. Kohl and F. Riesenfe1d, "Gas Purification", 3rd Ed. Gulf
Publishing Co., Houston (1979).
2l. A. R. Guzdar and A. C. Harvey, in Proc. Biomass-to-Methano1
Specialists Workshop, SERI/CP-234-1590, Tamarron, Colorado,
March 3-5, 1982.
22. R. Detman, "Factored Estimates for Western Coal Commercial Con-
Cepts", C. F. Braun & Co., Report FE 2240-5 (1976).
23. R. Detman, "Factored Estimates for Eastern Coal Commerica1 Con-
Cepts," C. F. Braun & Co., Report FE-2240-31 (1978).
24. B. Corni1s, J. Hibbe1, P. Ruprecht, R. Durfe1d, and J. Longoff,
Hydrocarbon Process., 60(1):149(1981).

25. M. S. Graboski, in Proc. Biomass-to-Methanol Specialists Workshop

SERI/CP-234-l590, Tamarron, Colorado, March 3-5, 1982.
26. L. A. Oster and H. J. Michaels, in Proc. Biomass-to-Methanol
Specialists Workshop, SERI/CP-234-l590, Tamarron, Colorado,
March 3-5, 1982.
27. W. A. Stevenson, in Proc. Biomass-to-Methanol Specialists
Workshop, SERI/CP-234-l590, Tamarron, Colorado, March 3-5,
28. T. B. Reed, M. Markson, and M. S. Graboski, in Proc. Biomass-
to-Methanol Specialists Workshop, SERI/CP-234-l590, Tamarron,
Colorado, March 3-5, 1982.


F. N. Lin and F. Pennella

Phillips Petroleum Company

Bartlesville, Oklahoma 74003


Typically, in the heterogeneously catalyzed hydrogenation of

carbon monoxide, copper-based catalysts are very selective for the
synthesis of methanol, while cobalt-based catalysts exhibit Fischer-
Tropsch activity with high selectivity to hydrocarbons. Mixed cop-
per-cobalt catalysts have been reported to show, in various degrees
[1-4], selectivity for higher alcohol formation. Of particular in-
terest are the copper-cobalt catalysts developed by Sugier and co-
workers [3,4] at the Institut Francais du Petro1e (IFP). Most of
these catalysts contain aluminum, chromium or zinc, and small amounts
of alkali, so that their composition corresponds to that of alkalized
conventional copper-based methanol synthesis catalysts modified by
the addition of cobalt. With these catalysts, high yields of higher
alcohols were obtained under methanol synthesis conditions, and, in
contrast with other copper-cobalt systems [1-2], there was no appre-
ciable methanation, and little [3] or moderate [4] formation of
higher hydrocarbons.

The distribution of products indicates [4] that these are not

bifunctional catalysts, comprising methanol formation and methanol
homologation, but that the higher alcohols are formed by a Fischer-
Tropsch mechanism. Little, however, is known about the nature of the
active sites [4]; in particular, it is not clear whether the higher
alcohols are formed over cobalt or copper sites, and whether copper
sites·active for methanol synthesis are modified by cobalt to produce
higher alcohols.

To elucidate the role played by cobalt in the modification of

the selectivity of methanol synthesis catalysts, a systematic study


has been carried out to determine the influence of cobalt on copper-

based catalysts. In this study, small incremental amounts of cobalt
were added to copper oxide/zinc oxide/alumina or to copper oxide/
chromium oxide/alumina catalysts, and the effects on catalytic be-
havior were determined.



Three commercial catalysts were used in this study. One, sup-

plied by UCI, had the composition CuO/ZnO/AlZ03, while the other two
had the composition CuO/CrZ03/AlZ03. They are labelled CuZnAlO.3'
CuCrO.Z6AlO.04' and CuCrl.ZAlO.95, according to the atom ratios of
the metals in the catalysts.

Cobalt was added to these catalysts by impregnation with aqueous

solutions of cobalt nitrate. A calculated amount of cobalt nitrate
was dissolved in ZO cc of distilled water; 10 g of catalyst (16/40
mesh) were added to the solution, and the bulk of the water was re-
moved by evaporation at 3l8°K; the catalyst was completely dried at
493°K for Z.4 hr, and calcined at 6Z3°K for 3.6 hr.

A CoO/ZnO!AlZ03 catalyst (CoO.3ZnAlO.3) was prepared by copre-

cipitation of cobalt and zinc carbonates following the procedure re-
ported [5] for the synthesis of CuZnAlO.3; ZnO/Alz03 (ZnAlO.3) was
prepared by analogous procedure.

Elemental analyses, BET surface areas and X-ray diffraction

analyses of the catalysts were obtained after calcination and after
reduction and catalytic testing. A Philips XRG-Z500 diffractometer
was used for the X-ray diffraction analyses. Average crystallite
sizes were measured by X-ray line broadening from the [110] reflec-
tion of ZnO, the [ZOZ] reflection of CuO and the [ZOOJ reflection
of Cu, using Scherrer's equation [6].

Catalytic Tests

The activity of the catalysts was evaluated in a fixed-bed,

single-pass, continuous-flow system equipped with a 1" i.d. tubular
reactor and with pressure, flow, and temperature controls. All ex-
periments were carried out with" 5 cc of catalyst dispersed in Z5 cc
of glass beads (3 mm). The catalysts were reduced in situ with hy-
drogen, flowing at 60 liters per hr, at 5Z3°K for 5 hr. A premixed
gas mixture containing hydrogen, carbon monoxide, carbon dioxide,
and nitrogen was then passed over the catalyst at 6.5 MPa to evaluate
its activity. For the experiments reported in Table 1 and Figure 1,
the gas composition was HZ/CO/COZ/NZ = 60/Z8/l0/Z, for all other ex-
periments the composition 60/Z8/Z/l0 was used.

Liquid products were condensed from the reactor effluent in a

knock-out container at 273°K and at reactor pressure. The pressure
was then reduced to atmospheric by a pressure-pilot-controlled re-
search valve and the effluent gases were analyzed with a multicolumn
Carle gas chromatograph (Model 197-S). The liquids collected in the
knock-out container were analyzed by gas chromatography using a
poropak Q column and an OV-lOl column.

Chemisorption Measurements

Adsorption isotherms were measured with a conventional constant

volume adsorption system. Each catalyst sample was reduced in a
flowing stream of hydrogen at 523°K for 5 hr and evacuated for 16 hr
at the same temperature before the chemisorption measurements. After
the isotherm was determined the sample was evacuated for 15 min and
a second isotherm was obtained to measure the weakly adsorbed gas.
The difference between the first and the second isotherm corresponds
to the strongly adsorbed gas.


The influence of cobalt on the catalytic behavior of the meth-

anol synthesis catalysts was determined by measuring initial (first
hour) rates of methanol formation. The results obtained with CuZn-
AlO.3 are shown in Table 1 and Figure 1. The methanol yield of the
base catalyst was 1.75 g/g-catalystjhr and selectivity to methanol
was greater than 99%. The addition to the catalyst of small amounts
of cobalt reduced the methanol synthesis activity dramatically, with-
out measurable changes in selectivity. For example, the addition of
0.3 wt% cobalt, corresponding to a cobalt surface coverage of 2%,
reduced the activity by nearly one order of magnitude. The decline
in activity continued as the cobalt content of the catalyst increased
to about 1%. With higher amounts of cobalt selectivity changes oc-
curred; methane and higher hydrocarbons were produced and both the
methanol and the higher alcohol yields increased. At low cobalt
concentration, there was an exponential relationship between the
amount of cobalt added and the catalyst deactivation (see Figure 2).
These results were obtained with the reactant gas mixture containing
10% C02; very similar results have been obtained with 2% C02 in the
feed [7].

The BET surface areas and the crystallite sizes of the catalysts
are listed in Table 2. The surface area of the calcined catalysts
decreased from about 40 m2 jg to 25 m2 jg as the cobalt content in-
creased from 0 to 8% (Figure 3). A similar trend is evident in the
surface areas of the tested catalyst, but with a greater scatter of
the datapoints. In the low cobalt concentration range (0-0.6 wt %),
where the loss of activity reached 98%, the surface area varied be-
tween 41 and 38 m2 jg. This small variation in surface area does not

Table 1. Activity of Cobalt Impregnated CuZnA1 0 • 3 Catalysts (at

(6.5 MPa, 543°K, and 8000 GHSV).

Yield, g/g/hr

Co Content Higher
Hydrocarbons Methanol
Wt % Alcohols

0.00 0.00 1.75 0.01

0.17 0.00 0.46 0.01
0.31 0.00 0.28 0.01
0.56 0.00 0.08 0.01
0.64 0.00 0.05 0.01
0.64 0.00 0.05 0.01
1.2 0.02 0.02 0.01
3.1 0.14 0.03 0.02
4.0 0.18 0.04 0.03
5.6 0.21 0.03 0.04
8.0 0.27 0.04 0.04

Table 2. Properties of Cobalt Impregnated CuZnA10.3 Catalyst a

Co Content °
Average Crlsta11ite Size! A
Wt % Area, m2 /g C0304 (Co) CuO (Cu) ZnO (ZnO)

0.00 38.7 (35.6) nd 125 (135) 133 (127)

0.17 41.2 (41.2) nd 125 (137) 142 (141)
0.31 40.6 (35.9) nd 128 (134) 123 (136)
0.56 (44.3)
0.64 38.5 (33.8) nd 110 (136) 148 (145)
0.64 38.4 (45.8) nd 108 (130) 125 (134)
1.20 35.0 (32.1) nd 112 (148) 146 (156)
3.10 30.2 (33.4) d (nd) 109 (130) 180 (183)
4.0 28.8 (23.8) d (nd) 115 (160) 220 (283)
5.6 26.7 (26.6) d (nd) 135 (140) 290 (290)
8.0 25.0 (27.3) 125 (nd) 140 (150) 240 (310)

aphases and values in parenthesis refer to the catalysts after re-

duction and synthesis tests; those outside parenthesis to the
calcined catalysts.
ndnot detected.
ddetected, but peak too small to determine crystallite size.



« 15
ci 1.0


\ ~
2.0 4.0
6.0 8.0

Figure 1. Activity of CuZnA10.3 as a function of the cobalt content.

account for the large decrease in the rate of methanol formation.

In the same concentration range, no cobalt species was detected by
X-ray diffraction (XRD) analysis, and the average crystallite sizes
of CuO, of Cu, and of ZnO showed no significant changes.

At higher cobalt concentrations, where the selectivity of the

catalyst changed and hydrocarbon formation was observed, the average
crystallite size of ZnO increased, an~ C0304 was detected by XRD
analysis. The peak positions in the diffraction pattern of the C0304


J: 0.5

;;: 0.1
~ 0.05

0.0 1!:------;;''=----:-'-:::-__-;:-L::--__~,-----l
o 2.0 4.0 6.0 B.O 1.0

Figure 2. Methanol yield over CuZnA10.3 as a function of cobalt



E u.i
<i N
300 iii
c:: w
<I: r
<.) j
~ 20 200 <I:
c:: r
(/) >-

O~------~~------~~------~~------~--------~ 0
o 2.0 4.0 6.0 8.0 10.0

Figure 3. Effects of cobalt on the surface area and crystallite

sizes of CuZnAl O• 3 •

were shifted to larger d-spacings, indicating the inclusion of zinc

or copper ions into the lattice.

Elemental analyses showed that no loss of cobalt occurred from

any of the catalysts during reduction and catalytic testing; also
there was no detectable loss of either copper or zinc. The ratios
of the crystalline CuD (or Cu) to the crystalline zinc oxide, deter-
mined from the XRD peak intensities, showed no variations outside
the range of the computational error.

Since the cobalt was added to the catalyst by impregnation with

aqueous solutions of cobalt nitrate, two control experiments were
performed to assess the effects of the nitrate ion and of the acidity
of the solution on the activity of the catalysts, and to determine
whether exchange occurred during impregnation between cobalt ions
and the copper ions, which are the precursors of the sites active for
methanol synthesis. In one experiment, the CuZnAIO.3 catalyst was
impregnated with a nitric acid solution containing the nitrate equiv-
alent of the solution used to prepare the 0.3% Co catalyst. The re-
sulting catalyst had a surface area of 37.3 m2 jg after calcination
and 32.5 m2 jg after reduction and synthesis testing, and it showed
no decrease in activity or selectivity with respect to the base cata-
lyst. In a second experiment the base catalyst was stirred for 30
min in a solution containing cobalt nitrate in the concentration
necessary to prepare a catalyst with 0.3% cobalt. The mixture was
filtered and the catalyst was washed with water. No copper was de-
tected in the filtrate and only traces of cobalt were found on the
catalyst. The resulting catalyst had the same activity and selec-

tivity as the base catalyst.

Synthesis tests were also carried out over zinc oxide-alumina

(ZnA10.3) and cobalt-zinc oxide-alumina (CoO.3ZnA10.3). The results
of these tests are shown in Table 3. Neither catalyst was active
after reduction in hydrogen at 523°K for 5 hr. When 1 wt % Cu was
added to the cobalt catalyst or when the catalyst was reduced at
773°K, hydrocarbons and alcohols were produced; the similarity in
the product yield and distribution indicates that the same catalytic
species is active in both cases. The copper apparently activates the
reduction of cobalt, without appreciable contribution to the conver-
sion of the synthesis gas. On the other hand, the addition of copper
to ZnA1 0 • 3 produced a highly selective catalyst for methanol synthe-

Hydrogen and carbon monoxide adsorption isotherm of the CuZn

A10.3 base catalyst and of the catalyst containing 0.64% cobalt were
measured at 216°K. The hydrogen adsorption isotherms of the two
catalysts were nearly identical, but CO adsorption, both weak and
strong, was lower on the catalyst with the cobalt (see Figure 4); the
strong co adsorption capacity of the base catalyst was 25 ~ mo1es/g
and that of the catalyst with cobalt 13 ~ moles/g. To determine the
contribution of the cobalt to the CO adsorption of the latter cata-
lyst, the CO adsorption isotherm of 0.6% Co on zinc oxide-alumina
was also measured. This sample was prepared by impregnation of
ZnA10.3 with an aqueous solution of cobalt nitrate. Because of the
difficulty in reducing cobalt at 523°K in the absence of copper, this
sample was pretreated with hydrogen at 773°K. The strong CO adsorp-
tion capacity of this sample was 2 ~ moles/g. There was no strong CO
adsorption on the zinc oxide-alumina. These results indicate that
in Co/CuZnA10.3 the contribution to the strong CO adsorption by the
cobalt deposited on the zinc oxide phase is small. The contribution
to the strong CO adsorption by cobalt occupying other surface sites
has not been determined. Therefore, the changes in CO chemisorption
produced by the cobalt cannot be correlated quantitatively with the
changes in methanol synthesis activity. However, the chemisorption
data in Figure 3 show unequivocally that the added cobalt diminished
the CO adsorption capacity of CuZnOA10.3.

The effects of cobalt addition on the catalytic behavior of the

copper-chromium oxide-alumina catalysts was studied only at low
cobalt concentrations. The results are given in Table 4. The addi-
tion of cobalt to the chromium-rich catalyst, CuCr1.2A10.9, caused
no significant changes in either activity or selectivity. With the
copper-rich catalyst, CuCrO.26A10.04' the methanol synthesis activity
declined with increasing addition of cobalt, but the decline was less
pronounced than with CuZnA10.3'


Table 3. Effects of Copper on the Activity of Cobalt-Zinc Oxide Catalysts

Yield, g/g-cat/hr a

Reduction Total
Catalyst Temperature (OK) Methanol Hydrocarbons

COO.3ZnA10.3 523 o o 0
Co O• 3 ZnA1 0 • 3 773 0.012 0.003 0.11
1% Cu/CoO.3ZnAlO.3b 523 0.014 0.006 0.09
ZnA1 0 • 3 523 o o 0
5% Cu/ZnA10.3b 523 0.18 0.18 0
aAt 543°K, 6.5 MFa, 8000 GHSV
bCopper was added by impregnation of the calcined catalyst with an aqueous solution of cupric z
nitrate. »


0 20
::t • • C •• ••

0 80 100

Figure 4. CO adsorption isotherms of CuZnAlO.3 (0) and of 0.6%

Co/CuZnAl O• 3 (').

The copper-zinc oxide-alumina has been studied in detail by

Klier and coworkers [8-10]. These authors, and others [11], have
concluded that the methanol synthesis activity of this catalyst is
associated with the presence of X-ray amorphous copper dissolved in
zinc oxide, that the strong CO adsorption at 296°K takes place on
these copper sites [10], and that alumina is not a necessary compo-
nent of the catalyst and acts only as a structural promoter [8].

The data presented in Table 1 clearly show that addition of

cobalt to copper-zinc oxide-alumina depresses the methanol synthesis
activity of the catalyst. In the range of cobalt concentrations in
which the deactivation of CuZnAlO.3 occurred without changes in se-
lectivity, there was neither a significant decrease in surface area
nor a significant change in the bulk properties of the catalyst.
Therefore, the decrease in methanol synthesis activity produced by
the addition of cobalt is attributable to surface phenomena, such as
inhibition of hydrogen or carbon monoxide adsorption, or inhibition
of the CO hydrogenation step. The decrease in the CO adsorption at
296°K caused by the addition of cobalt (Figure 4) suggests that the
cobalt deactivates the copper-zinc oxide sites. This deactivation
may take place through physical changes, such as the displacement of
the copper during calcination, or by disruption of the electronic
structure of the copper-oxide site by neighboring cobalt atoms. The
pronounced effects on the catalytic activity produced by small amounts
of cobalt indicate that the cobalt interacts quite selectively with
the catalytically active sites. For example, the addition of 29 ~
moles of cobalt per gram of catalyst (0.17 wt %) caused a 74% de-
crease in the rate of methanol formation. This amount of cobalt is
nearly equimolar with the strong CO adsorption capacity of CuZnAlO.3

Table 4. Activity of Cobalt Impregnated Copper-Chromium-Alumina

Catalysts (at 6.5 MPa, 543°K, and 8000 GHSV).

Catalyst Wt % Co Yield of MeOH, g/g-cat/hr

CuCrl.2AlO.9 0 0.37
CuCrl.2AlO.9 0.1 0.36
CuCrl.2Al O.9 0.45 0.40
CuCrl.2AlO.9 1.80 0.35
CuCrO.26 A1 0.04 0 0.26
CuCrO.26AlO.04 0.27 0.16
CuCrO.26AlO.04 0.55 0.11
CuCrO.26AlO.04 1.1 0.036

(25 ~ moles/g) and corresponds to a 1% monolayer surface coverage.

These results lead to the conclusion that the added cobalt deacti-
vates CuZnAlO.3, but does not modify the copper sites active for
methanol synthesis. At concentrations of cobalt greater than about
1 ",t %, activity and selectivity changes were observed; hydrocarbons
weY'e formed and the yields of alcohols increased with greater selec-
tivity to higher alcohols than the base catalyst. The distribution
of products is similar to that observed over COO.3ZnA10.3 (Table 3),
so that these new activity and selectivity patterns can be attri-
buted to the development of the catalytic activity of cobalt.

Comparison of the results obtained with CuZnAlO.3 with those

obtained with CuCrl.2AlO.9 shows a sharp difference in the dependence
of the methanol synthesis activity of the two catalysts on cobalt.
The two catalysts contain similar amounts of copper, but the Cu-Zn-Al
catalyst was strongly poisoned by the cobalt, while the Cu-Cr-Al
catalyst was not affected at equivalent Co concentrations. At high
concentrations of Cu, i.e. with CuCrO.26AlO.04, deactivation of the
catalyst occurred. These results indicate that the active sites are
poisoned by the cobalt on both catalysts. However, on the Cu-Zn
catalyst the cobalt interacts preferentially with the active sites,
whereas on the Cu-Cr catalyst it interacts preferentially with in-
active Cr species.

The purpose of this work was to determine whether copper sites
active for methanol synthesis are modified by cobalt to produce
higher alcohols. The results obtained indicate that on either cop-
per-zinc or copper-chromium-based catalysts, cobalt deactivates the
sites and does not alter their selectivity. The formation of higher
alcohols and hydrocarbons observed on CuZnAlO.3 at high concentra-
tions of cobalt can be attributed to the formation of cobalt ensem-
bles and not to modifications of the methanol synthesis sites.


1. R. Taylor, J. Chem. Soc. (London), 1429 (1934).

2. M. K. Zaman Khan, C. H. Yang, and A. C. Oblad, Prel2rintsz Div.
Fuel Sci. z ACS, 22(2):138 (1977) •
3. A. Sugier and E. Freund, U.S. Patent 4,122,110 (1978) and German
Offen. 2,748,097 (1978); assigned to Institu Francais du
4. Ph. Courty, D. Durand, E. Freund, and A. Sugier, Symposium on
Catalytic Reaction of One Carbon Molecules, Bruges, June 1982.
5. T. D. Casey and G. M. Chapman, U.S. Patent 3,790,505 (1974);
assigned to Catalysts and Chemicals, Inc.
6. A. Guinier, "X-Ray Diffraction," W. H. Freeman, New York, (1963),
p 122.
7. D. J. Elliott, private communication.
8. S. Mehta, G. W. Simmons, K. Klier, and R. G. Herman, J. Catal.,
57:339 (1979).
9. K. Klier, V. Chatikavanij, R. G. Herman, and G. W. Simmons,
J. Catal., 74:343 (1982); and references therein.
10. K. Klier and R. G. Herman, Final Technical Report to DOE,
FE-3l77-F (1980).
11. s. V. Ketchik, T. P. Minyukova, L. I. Kuznetsova, L. M. Plya-
sova, T. M. Yurieva, and C. K. Boreskov, React. Kinet. Cat.
Letter, 19:345 (1982); ibid. 19:355 (1982).


G. B. Atkinson, E. G. Baglin, L. J. Nicks, and D. J. Bauer

u.S. Department of the Interior

Bureau of Mines, Reno Research Center
1605 Evans Avenue
Reno, Nevada 89512


The Bureau of Mines, Reno Research Center, began a catalyst

research project in the mid-seventies to investigate improving the
efficiency of catalysts containing critical and strategic metals.
The initial years were spent gathering equipment and developing meth-
ods for preparation, physical characterization, and testing activity
of experimental catalysts. Early research concentrated on the cat-
alytic properties of rare-earth elements for oxidation reactions [1].
In addition to the catalyst research, another group at Reno was in-
vestigating the hydrogen absorption properties of intermetallic com-
pounds of the type LaNis [2]. Since hydrogen storage alloys func-
tioned by breaking the hydrogen bond and storing atomic hydrogen
[3,4], they were logical candidates for study as hydrogenation cata-
lysts. Synthesis of methane was the first reaction studied.

During testing of cerium-nickel intermetallic compounds as

catalysts for the methanation reaction, at 10 atm pressure and 400°C,
conversion of 3H2:CO synthesis gas to methane increased from a few
percent to complete conversion during the first several hours. The
alloy CeNis was apparently activated by exposure to the synthesis
gas. Further research on CeNis showed that a two-step treatment was
best for developing active catalysts with high surface area. The
initial step involved treatment of CeNis with synthesis gas at about
400°C to destroy the intermetallic compound b oxidizing cerium to
Ce02. This resulted in a surface area of 3 m /g. The second step
involved treatment of the mixture with synthesis gas at l7SoC. The
resulting catalyst was composed of nickel dispersed in Ce02 with a
surface area of 40 m2 /g.


Catalysts prepared from CeNiS had specific activity for meth-

anation that was five times greater than commercial nickel on alum-
ina catalysts and were equal to or better than commercial catalysts
with regard to sulfur tolerance. The catalysts were many times
more active than Ce02/SNi catalysts prepared by conventional copre-
cipitation techniques with identical composition and X-ray diffrac-
tion patterns.

Other rare earths and reactive metals were substituted for

cerium in CeNis and the alloys were tested for surface area devel-
opment and methanation activity. Rare earths, such as Y, La, Pr,
Nd, Sm, and Eu, and other metals, such as Ca, Zr, Hf, Th, and U, were
more or less successfully substituted for cerium. The metals Ti, V,
Ta, Nb, and Al were unsuccessful because temperatures higher than
600°C were necessary to oxidize the metal. The high temperature for
oxidation promoted sintering, which decreased surface area devel-

One exception to the two-step treatment for high surface area

development was ThNiS alloy. Oxidation of the other alloys with
synthesis gas at 400°C produced total surface areas of only 3 m2 /g,
but this treatment produced total surface areas of 2S m2 /g from
ThNiS alloys. Thorium-containing alloys reacted differently during
oxidation and produced high surface areas from a single-step treat-

Properties of intermetallic compounds as catalyst precursors

have been reported in patents and publications by the Bureau of
Mines and other researchers [S-23].

The high activity for methanation demonstrated by catalysts

prepared from intermetallic compounds suggested that catalysts pro-
duced from metal alloys should be investigated for other hydrogen-
ation reactions. During screening experiments on metal alloy cata-
lysts for hydrogenation reactions, alloys containing copper exhibited
favorable activity for the synthesis of methanol from hydrogen and
carbon monoxide. Detailed experiments were initiated to investigate
the activity of different metal combinations for methanol synthesis
and to determine methods for treating the alloys to develop surface

Modern industrial methanol production is based on the conversion

of pressurized synthesis gas containing hydrogen, carbon monoxide,
and carbon dioxide in the presence of a metallic catalyst. Some
high-pressure plants (2S0-3S0 atm) still exist, but these are being
replaced by larger, more energy efficient low-pressure plants (SO-
100 atm). The industrial low-pressure plants use catalysts based
on the compositions Cu/ZnO/Cr203 or Cu/ZnO/A1203 [24]. Therefore,
alloys prepared for screening experiments for the methanol reaction
contained either copper or zinc.


Alloy Preparation

Many binary and a few tertiary alloys, mostly intermetallic

compounds, were prepared for screening tests. For alloys containing
copper, the metals were weighed, mixed, and consolidated by arc-
melting on a water cooled copper hearth and in a helium atmosphere.
Alloys containing zinc could not be arc-melted because of the high
vapor pressure of molten zinc; instead, they were prepared by weigh-
ing the stoichiometric quantities of the metals into quartz tubes,
which were evacuated and sealed. The tubes were heated to a temper-
ature slightly above the melting point of the desired alloy for 0.75
to 3.5 hr. The alloys were heat aged for 16 to 40 hr at approxi-
mately 20 to 50°C below their melting points. If the melting point
of the alloy was above 1,000°C, the metal mixture was not melted
but was heated to 1,000°C and aged for several days. All alloys
tested were in the as-prepared form, and the formulas given represent
the alloy composition and not necessarily single-phase material.

Several alloys containing thorium or uranium were prepared.

Caution should be observed in handling and storing these alloys be-
cause of their radioactive nature and because they tend to disinte-
grate with the evolution of heat when they are stored at ambient
conditions. The pyrophoric nature was especially noticeable for
Th2Cu and ThCu2.

For comparison of catalytic activity, a Th02/5Cu sample was pre-

pared by coprecipitation of the stoichiometric amounts of thorium
and copper nitrates by pH adjustment using Na2C03 solution. The
precipitate was filtered, washed with water, dried at 110°C, and cal-
cined at 350°C. Reduction of the copper oxides occurred in the syn-
thesis reactor. A commercial Cu/ZnO/AI203 catalyst was also evalu-
ated in comparative tests.


The alloys were ground to minus 25 plus 80 mesh and were indi-
vidually tested to develop surface area by treating with different
oxidizing gases at elevated temperatures. The gas mixture must have
sufficient oxidizing power to convert the reactive metal to its
oxide. Frequently used gas mixtures included air, air saturated with
water, carbon dioxide, hydrogen saturated with water, and hydrogen
and carbon monoxide. Th2Cu and ThCu2 alloy buttons disintegrated
when the thorium was oxidized to thoria in air at ambient conditions.
ThCu3.6 partially disintegrated under these conditions.

Reactor Design and Testing

For screening experiments, 0.2-ml samples of activated catalyst


or ground alloy were supported between plugs of Pyrexa wool in a

1/4-inch-ID stainless steel continuous flow reactor. The reactor
pressure was controlled with a heated back pressure regulator. All
gas lines were contained in an oven at 200°C to prevent condensation
of reaction products. Helium was used to pressurize and flush the
reactor. After out-gassing, the helium was replaced with a commer-
cial analyzed synthesis gas mixture. Inlet flow was maintained by
a fine-metering valve. The temperature was increased to 220°C and
held until reproducible analysis of the exit gas was obtained. The
temperature was increased in 10 to 20°C increments and the exit gases
were analyzed by gas chromatography to obtain conversion data. The
experiment was terminated when the amount of methanol produced start-
ed to decrease at higher temperatures because of the effect of equi-

Screening experiments were made with hydrogen to carbon monoxide

synthesis gas mole ratios of 16:1 at 60 atm pressure and a synthesis
gas space velocity (GHSV) of about 31,000 hr- l • The high ratio of
hydrogen to carbon monoxide enabled product analysis by preventing
overloading of the chromatographic columns with carbon monoxide from
sample loops at 60 atm pressure. Later modifications of the system
to decrease the pressure in the sample loops to atmospheric pressure
allowed experiments to be conducted with a synthesis gas ratio of
2. 5H2: CO •

Reactor exit gas was analyzed with a gas chromatograph using

Chromo sorb 102 and 5A molecular sieve columns in a series-bypass
configuration. A thermal conductivity detector and integrating re-
corder completed the analytical system. Since the gas chromatograph
was unable to analyze hydrogen, the results are reported as mole
percent methanol in the reactor exit gas and exclude hydrogen. Liq-
uids were sometimes condensed and collected in a dry ice-acetone
trap. The liquid product was analyzed by chromatography with a Chro-
mosorb 103 column and a flame ionization detector. Surface areas
were measured by the standard gravimetric BET nitrogen adsorption

A larger sample of catalyst ThCu6, 1.0 ml, was tested in the

reactor system shown in Figure 1. Bottled analyzed 2.4H2:CO syn-
thesis gas containing 3 vol % nitrogen as an inert internal analyt-
ical standard was used. Before entering the reactor, the synthesis
gas was purified by heating to 350°C to decompose nickel carbonyl
and passed through an activated charcoal filter to remove other im-
purities. The apparatus included a steady state, single pass, flow

aReference to specific materials does not imply endorsement by the

Bureau of Mines.

350· C

u Thermocouple


Synthesis gas trap
and helium


Figure 1. Reactor system for testing methanol activity.

through, fixed bed reactor with a diaphragm flow controller, mass

flowmeter, and a back pressure regulator. Condensables in the exit
gas were contained in a trap maintained at -IODC and reactor pres-
sure. The stripped gases were analyzed by a gas chromatograph capa-
ble of analyzing hydrogen. The sample temperature was monitored
with a O.lZ5-inch-OD stainless steel sheathed thermocouple placed
in the center of the sample bed. A constant temperature was main-
tained in the O.5-inch-OD stainless steel reactor by using a 3-inch-
diam by 8-inch-long aluminum cylinder as a heat sink. Holes were
drilled in the aluminum cylinder to accommodate the reactor, a con-
trol thermocouple, and two O.5-inch-diam by 6-inch-long cartridge


Alloys containing a reactive metal and copper, zinc, and chro-

mium were prepared and treated with different gas mixtures to inves-
tigate the development of surface area. Samples that developed
surface area were tested for methanol synthesis activity. If the
copper or zinc was oxidized during surface area development, it was
reduced in the reactor before testing. Table 1 shows a listing of
the data obtained during screening representative catalyst samples
with the greatest surface area for methanol activity. Included for
comparison are results from testing the commercial Cu/ZnO/AlZ03 cata-
lyst. The second column shows the gas mixture and temperature used

Table l. Results from Screening Experiments to Determine

Hethanol Activity of Catalysts

Catalyst Activation Total area CH 30H at
Treatment mZ/g mol % maximum, °c

Cu/ZnO/A1Z03 4Z 13.0 3Z0

ThCu5 Air/400°C Z6.0 Z6.0 Z80
ThCu5 Air/HZO/400°C Z6.Z Z6.5 300
ThCu5 COZ/450°C 33.6 Z4.9 300
ThCu5 HZ/HZO/400°C ZO.5 17 .8 300
ThCuCr COZ/450°C 3Z.4 l6.Z 300
HfCu5 3HZ/CO/400°C Z9.8 9.6 3Z0
UCu5 3HZ/CO/400°C 16.7 7.Z 3Z0
CeCuz HZ/HzO/400°C 9.9 7.0 340
ZrCu5 3HZ/CO/400°C lZ.O 6.4 3Z0
CeCu5 Air/HzO/400°C 13.0 6.1 300
ThZnz COZ/450°C Z.Z 3.7 360
ThZn4 Air/HzO/400°C 4.9 3.4 360
CeCuZn COZ/450°C 9.6 Z.l 380
CeZn COZ/450°C Z1.l .9 380
CeZnz COZ/450°C 13.1 .6 400

to treat the alloy for surface area development. The total surface
area was measured after methanol testing. Results are reported as
the maximum mole percent methanol in the product gas and exclude
hydrogen. The temperature at which each maximum was obtained is

The surface area developed and the methanol produced were low
for catalysts prepared with zinc as the active catalyst component.
Maximum methanol production was obtained at temperatures of 360 to
400°C. Copper-containing catalysts were more active than the zinc-
containing catalysts and maximum methanol production was obtained
at temperatures of Z80 to 350°C. Maximum production of methanol for
some of the alloy-based catalysts was greater than the commercial
catalyst and was obtained at lower temperatures. The alloy ThCu5
developed surface area by several different activation methods. Ox-
idation with air at 400°C produced the most active catalyst. A
series of thorium-copper alloys was prepared and investigated to
determine the best composition.

The binary phase diagram for the thorium-copper system shows

the existence of four intermetallic compounds--ThZCu, ThCuZ, ThCu3.6,
and ThCu6' These intermetallics and materials of nominal composi-
tion ThCu5, ThCuCr, thorium containing 5% Cu, ThOZ/5Cu coprecipitate,

and a commercial Cu/ZnO/A1203 were tested for methanol activity.

The intermetallics Th2Cu and ThCu2 slowly disintegrated in air

at ambient conditions because the thorium oxidized to thoria.
ThCu3.6 partially disintegrated under these conditions. No further
activation was necessary for these catalysts. The intact portion
of ThCu3.6 and ThCuS, ThCu6, Th/S % Cu, and ThCuCr alloys were
crushed to minus 25 plus 80 mesh. The granules were activated by
reaction with air at 400°C. Oxidation of thorium during the air ac-
tivation procedure resulted in expansion of the catalyst ranging
from a few percent to as much as 40%.

All catalysts were reduced with synthesis gas in the methanol

synthesis reactor before testing 0.2 ml samples. Reactor conditions
included l6H2:CO synthesis gas at 60 atm and at a space velocity of
31,000 hr- l •

Figure 2 shows the production of methanol as a function of tem-

perature for the ThCu x alloys in which the thorium was oxidized to
thoria in air at ambient conditions. Temperature profiles for Th02/
SCu, prepared by coprecipitation, and Cu/ZnO/A1203 are included for
comparison. Figure 3 shows the temperature profiles for the alloys
that were activated in air at 400°C. ThCu3.6 was only partially de-
graded at ambient conditions and is included in each figure. The
ordinate of the temperature profiles represents mole percent methanol
in the exit gas mixture, excluding the hydrogen, which could not be
measured. This is approximately equal to the mole percent CO con-
verted to methanol. The equilibrium curve was calculated after Ewell
[25] and shows the maximum possible conversion of carbon monoxide
to methanol at 60 atm pressure from 16H2:CO synthesis gas. The ex-
perimental curves possess maxima because of the inhibiting effects
of equilibrium at higher temperatures.

The catalysts produced from alloy precursors were many times

more active than the commercial and coprecipitated materials.
ThCu3.6 and ThCu6 showed the highest activity. The total surface
areas of the catalysts after testing are included in Figure 3. Areas
as high as 40 m2 /g were attained. X-Ray diffraction showed the air
activated catalysts contained thoria, copper, and copper oxides.
The used alloy catalysts contained thoria and copper, and also chro-
mium in the case of the ThCuCr sample. The copper oxides were re-
duced to copper by the synthesis gas. Particle size degradation
occurred during the test of air-activated samples. It should be
pointed out that the commercial Cu/ZnO/AI203 catalyst was evaluated
under conditions very different from those of an industrial opera-
tion, where carbon dioxide is often added to increase the carbon
content of the gases and to maintain catalytic activity and where
synthesis gas of high hydrogen content is not used.)

An analysis of the reactor offgases for a typical catalyst is



........... Th 2Cu, 27.9 m 2/g

- - - ThCu2' 38.2 m2 /g
- - - - - ThCu3.6' 40.7 m2/g
------- Cu/ZnO/AI~3, 42.0 m2/g
~ 50 - - - Th02/5Cu coprecipitate,
................ 20.4 m2 /g
C\I ......'
co 40
.' .
/''''' / / - - .......... ~'\
)( / \
~ 30
....... ------ .....
. ....
" ...
... , ' .------ ----------- ........
Figure 2. Methanol production as a function of temperature. Syn-
thesis gas: 16H2:CO, 60 atm pressure, 31,000 GHSV.

shown in Table 2. The amount and number of side products formed

increased with temperature. A small quantity of dimethyl ether was
present at temperatures greater than 300°C. In some cases, the off-
gasses were condensed and analyzed. Liquid product purity was more
than 99% methanol, with ethanol as the principal impurity. Higher
alcohols formed in lesser amounts.

Modification of the sample loops allowed the synthesis gas ratio

to be decreased from 16H2:CO to 2.5H2:CO without overloading the
chromatographic columns. The granular thorium-copper alloys were
tested without pretreatment. The 2.5H2:CO synthesis gas at 60 atm
reacted with the alloys, producing Th02 and copper metal. Based on
the volume of the original alloy, the space velocity was 31,000 hr-1 ,
but because of expansion by the samples during oxidation, the actual
space velocity was less. Figure 4 shows temperature profiles of
alloys between the composition ThCu3.6 to ThCu9 and ThCuCr. The
ThCu6 alloy produced the largest amount of methanol at 31 mol % meth-
anol in the exit gas (excluding hydrogen) at 260°C. Liquid product
purity remained at more than 99% methanol, with ethanol the prin-
cipal impurity.

Results from testing small samples of ThCux alloys (0.2 m1) for
methanol activity showed that ThCu6 was the best composition, and


ThCu3.6' 35.5 m2/g

60 - - - ThCu5, 26.0 m 2/g
•••••••••••• ThCu6' 30.6 m2/g

------- Th/5% Cu, 3.8 m 2/g
~ 50
- - - ThCuCr, 18.0 m 2/g
CJI 40

~ 30
~ 20

Figure 3. Methanol production as a function of temperature. Syn-
thesis gas: l6H2:CO, 60 atm pressure, 31,000 GHSV.

treatment with 2.5H2:CO synthesis was satisfactory for activation.

Larger samples of ThCu6 (1.0 ml) were tested in the reactor system
shown in Figure 1. Samples were evaluated to determine the long-
term stability of the catalyst and to determine the effect of syn-
thesis gas pressure on methanol production. The stability of a
ThCu6 alloy catalyst (9B-120) was tested at the best temperature of
260°C, which was determined from analyzing the data collected during
testing small samples of ThCu6 activated with 2.5H2:CO synthesis gas.
One cubic centimeter of ground ThCu6 weighing 5.3 g was loaded into
the 1/2-inch-OD stainless steel reactor. Synthesis gas with a 2.4H2:
CO ratio was flowing at an initial space velocity of 21,360 hr- l ,
and the reactor was slowly heated to 260°C. The slow increase in
temperature was necessary to prevent a rapid temperature increase
due to the exothermic oxidation reaction of synthesis gas with ThCu6'

In the following discussion and analysis, "exit gas" means

total gas and vapor leaving the reactor; "stripped gas" is gas after
removal of condensibles. Gas analysis is in volume %, and liquid
analysis is in weight %. Methanol in the exit gas was determined
by weighing the fluid contained in the liquid trap cooled to -10°C.
Gas analysis includes the hydrogen.

Liquid product and stripped gas were periodically sampled and


Table 2. Gaseous Product Analysis as a Function of Temperature

from the Synthesis of Methanol over ThCu3.6, Preactivated
in Air at 400°C, l6H2:CO, 60 atm Pressure, 31,000 GHSV

Temperature Mole %, excluding H2

°c CO CH4 CO2 C2H2 H2 O CH30H ( CH3)20

220 83.52 0.24 0.08 16.16
240 67.22 0.60 0.28 31.89
260 55.93 1.24 1.07 0.06 41.70
270 52.26 1.99 1.58 0.08 0.45 43.54
280 49.42 2.85 2.46 0.l3 1.11 44.02
300 54.44 5.49 3.65 0.17 1.91 34.04 0.31

analyzed during 475 hours of operation. Figure 5 shows the experi-

mental results. After an initial conditioning period of approxi-
mately 90 hours, the catalyst sample showed good stability for meth-
anol production for 475 hours. The production of methane and carbon
dioxide byproducts decreased. The total condensed liquid contained
more than 98% methanol; the major impurities were ethanol and water.
After the experiment was completed, examination of the catalyst
showed that the synthesis gas had oxidized the thorium and increased
the catalyst volume from 1.0 to 2.25 ml and the weight from 5.3 to
6.0 g •. The volume increase resulted in a decrease in space velocity
from 21,360 to 9,500 hr-l. The used sample contained 1.0 wt %
carbon but no obvious carbon deposit. X-Ray diffraction showed only
Th02 and copper.

Another sample of ThCu6 (9B-173), which was slightly more

active, was tested to determine the conversion to methanol as a
function of temperature and synthesis gas pressures of 20, 40, and
60 atm. For comparative purposes, a commercial Cu/ZnO/A1203 cata-
lyst was tested at identical experimental conditions at 60 atm pres-
sure. Results from these experiments are shown in Figure 6. At
maximum conversion to methanol, the ThCu6 catalyst produced more than
10 times as much methanol as the Cu/ZnO/A1203 catalyst. At 250°C
and 60 atm pressure, ThCu6 produced more than 15% methanol in the
exit gas, and Cu/ZnO/A1203 yielded less than 0.5%. Even at 20 atm
pressure and 250°C, the ThCu6 produced more than 4% methanol. All
data were obtained after the catalyst had been held at 240°C under
methanol production conditions and at 60 atm for more than 90 hours.
The equilibrium curve was calculated after Ewell [25].

Analyses of stripped gas and condensed liquid were obtained at

periodic intervals during each experiment. Typical results are
reported in Tables 3 and 4. The data show that the purity of the
methanol produced was comparable for ThCu6 catalysts and Cu/ZnO/A1203

------ ThCU3.6. 35.6 m 2/g

- - - ThCU5. 23.2 m2/g
,.e ••••••••••••• ThCu6. 14.8 m2/g

"0 30
.............................. ------- ThCU9. 15.0 m2/g

- - - ThCuCr. 23.7 m2/g
N .....
,.,."..""" ,.,---........"
c: A--- __ " "-
'0 /"" / -~~,-
.2 20


, /

I "I ,

/,,..,,,, /
I- 10

O~ _ _ _ _ _ _ _L-____________L-____________L -__________-...J
200 250 300 350 400

Figure 4. Methanol production versus temperature for alloys with-

out pretreatment. Synthesis gas: 2.5H2:CO, 60 atm pres-
sure, 31,000 GHSV.

catalyst. A comparison of conversion versus temperature at identi-

cal reactor conditions is shown in Table 5. As an aid in comparing
with other results, the data are presented in three different ways:
g CH30H/ml of catalyst/hr, vol % CH30H in the exit gas, and % CO
converted to CH30H.


A new catalyst for the synthesis of methanol from hydrogen and

carbon monoxide was developed. Preparation of the catalyst was
accomplished by melting thorium and copper metal to form an interme-
tallic compound. The alloy was crushed to a convenient particle
size and exposed to an oxidizing gas at temperatures between 200°C
and 400°C. This oxidized the thorium to thoria and yielded a high
surface area catalyst containing copper interspersed with thoria.
The best composition was ThCu6 activated in situ with H2-CO synthe-
sis gas at 60 atm pressure. Catalysts prepared from ThCu6 were on
the order of 10 times more active than a commercial Cu/ZnO/A1203
catalyst or a coprecipitated Th02/Cu catalyst of similar composition.
A continuous decrease in the formation of the gaseous impurities
methane and carbon dioxide was observed with time. The catalyst ex-
hibited good stability for methanol production for 475 hours.

Table 3. Analysis of Typical Stripped Gas at 60 atm

Temperature Time Vol %

°c hr H2 CO N2 CO2 CH4
9B-173 240 93 70.28 25.08 4.27 0.22 0.15
9B-120 260 475 68.96 23.95 4.32 1.32 1.46
Cu/ZnO/A1 203 240 90 68.20 28.77 3.00 0.03

Table 4. Analysis of Typical Liquid Product at 60 atm 2.4H2:COa

Temperature Time Wt %
°c hr CH30H H20 C2H50H >C2H50Hb

9B-173 240 93 99.01 0.18 0.44 0.37

9B-120 260 475 98.05 0.62 0.80 0.53
Cu/ZnO/Al203 240 90 98.89 0.83 0.20 0.09
aThe volatility of dimethyl ether prevented its determination and
was estimated at <1%.
bTotal alcohols with molecular weight higher than ethanol.

Table 5. Comparison of Conversion versus Temperature at 60 atm

2.4H2:CO, -22,000 GHSV

% CO
Temperature g CH30H/ % CH30H in converted to
°c ml/hr exit gas CH30H

180 1.04 3.75 13.0
200 1.59 5.87 19.4
210 1.91 7.19 23.2
230 2.77 11.06 33.6
250 3.55 15.46 44.5

240 0.10 0.32 1.1
280 0.24 0.82 2.8
300 0.37 1.27 4.3
320 0.52 1.84 6.2
340 0.49 1.77 6.0

15 4#-

CHsOH en

C) 0
!: Q.
X !!:
1&1 II::
~ II)

0 ~

Figure 5. Catalytic stability of ThCu6 (9B-120). Synthesis gas:
H2:CO = 2.4, 60 atm pressure, 9,500 GHSV, and at 260°C.



cr Cu/ZnO/AI20S
~ 60 a~t;.;.m_ _......
300 340

Figure 6. Methanol production versus temperature for ThCu6 (9B-l73).

Synthesis gas: H2:CO = 2.4, 22,000 GHSV.


1. L. J. Nicks and D. J. Mac Donald, "Catalytic Activity of Rare-

Earth Oxides for the Oxidation of Hydrogen," BuMines RI 7841,
9 pp (1973).
2. R. A. Guidotti, G. B. Atkinson, and M. M. Wong, J. Less Common
Metals, 52(3):13 (1977).
3. F. A. Kuijpers and B. O. Loopstra, J. Phys. Colloq., 32:657
(1971) •
4. J. H. N. Van Vucht, F. A. Kuijpers, and H. C. A. M. Bruning,
Phillips Res. Repts., 25:133 (1970).

5. G. B. Atkinson and L. J. Nicks, U.S. Patent 4,071,473 (Jan. 31,

1978); assigned to U.S. Department of the Interior.
6. G. B. Atkinson and L. J. Nicks, U.S. Patent 4,256,653 (March
17, 1981); assigned to U.S. Department of the Interior.
7. G. B. Atkinson, L. J. Nicks, and D. J. Bauer, U.S. Patent
4,287,095 (Sept. 1, 1981); assigned to U.S. Department of
the Interior.
8. G. B. Atkinson, L. J. Nicks, and D. J. Bauer, U.S. Patent
4,301,032 (Nov. 17, 1981); assigned to U.S. Department of
the Interior.
9. G. B. Atkinson, L. J. Nicks, E. G. Bag1in, and D. J. Bauer,
"Hydrogenation Catalysts from Intermeta11ic Compounds," Bu-
Mines RI 8631, 8 pp (1982).
10. G. B. Atkinson and L. J. Nicks, J. Cata1., 46:417 (1977).
11. E. G. Bag1in, G. B. Atkinson, and L. J. Nicks, U.S. Patent
4,181,630 (Jan. 1, 1980); assigned to U.S. Department of
the Interior.
12. E. G. Bag1in, G. B. Atkinson, and L. J. Nicks, Ind. Eng. Chern.,
Prod. Res. Dev., 20(3):87 (1981).
13. V. T. Coon, T. Takeshita, W. E. Wallace, and R. S. Craig, J.
Phys. Chern., 80:1878 (1976). -
14. V. T. Coon, W. E. Wallace, and R. S. Craig, in "The Rare Earths
in Moderr Science and Technology," Vol. 1, ed. by G. J.
McCarthy and J. J. Rhyne, P1inum Press, New York, pp. 93-98
15. A. E1attar, T. Takeshita, W. E. Wallace, and R. S. Craig,
Science, 196:1093 (1977).
16. A. E1attar and W. E. Wallace, in "The Rare Earths in Modern Sci-
ence and Technology," Vol. 2, ed. by G. J. McCarthy, J. J.
Rhyne, and G. B. Silver, Plenum Press, New York, pp 533-537
(1980) •
17. A. E1attar, W. E. Wallace, and R. S. Craig, Adv. Chern. Ser.,
178:7 (1979).
18. A. E1attar, W. E. Wallace, and R. S. Craig, in "The Rare Earths
in Modern Science and Technology," Vol. 1, ed. by G. J. Mc-
Carthy and J. J. Rhyne, Plenum Press, New York, pp 87-92
19. H. Imamura and W. E. Wallace, J. Phys. Chern., 83:3261 (1979).
20. H. Imamura and W. E. Wallace, J. Phys. Chern., 83:2009 (1979).
21. H. Imamura and W. E. Wallace, J. Cata1., 65:127 (1980).
22. T. Takeshita, W. E. Wallace, and R. S. Craig, J. Cata1., 44:
236 (1976).
23. W. E. Wallace, J. France, and A. Shamsi, in "The Rare Earths in
Modern Science and Technology," Vol. 3, ed. by G. J. Mc-
Carthy, H. B. Silver, and J. J. Rhyne, Plenum Press, New
York, pp 561-568 (1982).

24. L. E. Wade, R. B. Genge1bach, J. L. Trumb1ey, and W. L. Ha11-

baur, in "Kirk-Othmer Encyclopedia of Chemical Technology,"
Vol. 15, 3rd Ed., ed. by M. Grayson, John Wiley & Sons,
New York, p 398 (1981).
25. R. H. Ewell, Ind. Eng. Chern., 32:147 (1940).

John F. Knifton a , Robert A. Grigsby, Jr.a and

Sheldon Herbstmanb

aTexaco Chemical Company, Austin, Texas 78761

bTexaco Research Center, Beacon, New York 12508


It is estimated that, by 1990, 55 percent of the world's

gasoline market will no longer contain lead antiknock compounds [1].
This shift to unleaded and low-lead gasolines will raise clear
octane requirements by about two octane numbers in the United States
and Europe, and while this increase could be satisfied by higher-
severity processing in the refinery, octane improvers will become
an increasingly attractive alternative. Methyl tertiary butyl
ether (MTBE) has been tagged as the favored octane booster, but its
production will be limited by the availability of suitable iso-
butylene streams [2,3]. Other major competitors for the octane
improver market include gasoline-grade tertiary butyl alcohol
(GTBA) [4,5], methanol [5,6], ethanol [7-9], and toluene [1].
Ethyl fuel [10], comprising mixtures of ethanol and methyl acetate,
has also been tested. The cited benefits [4,8] of using such blend-
ing oxygenated compounds include a) deferred capital requirements
for increasing gasoline octane, b) improved ability to cope with
more severe lead restrictions, c) improved front end octane per-
formance, and d) reduced exhaust emissions of air pollutants.

As part of our continuing research into the selective produc-

tion of aliphatic oxygenates from synthesis gas [11-13], we describe
in this paper a process for generating CI-C3 alcohols and their
acetate esters (eq. 1) directly from CO/H2 using the patented con-
cept of ruthenium "melt" catalysis [14]. This very flexible process
can be used to generate a diverse range of oxygenates that includes
methanol, ethanol, propanols, their acetate esters, as well as
mixtures thereof. In continuous processing, the scope of this
catalysis is exemplified by the triruthenium dodecacarbonyl-tetra-


butylphosphonium iodide combination. Some preliminary costing data

and a process schematic have been developed. Fuel performance data
are discussed for two typical Cl-C3 alcohol and Cl-C3 alcohol-ester
product combinations.


Prior publications featuring the generation of lower alkanols

from synthesis gas have in many cases focused on the use of modified
Fischer-Tropsch catalysts. The recent patents [15,16] issued to
the Institut Francais du Petrole (IFP) in particular, describe a
series of Cu-Co-multimetallic catalysts that produce normal alcohols
in >95% selectivity, and ethanol plus higher alcohols in yields ex-
ceeding 70%. Union Carbide likewise generates ethanol, plus other
oxygenates, by passage of CO/HZ over rhodium-on-silica gel catalysts
[17,18]. Other silica and lanthanum oxide-supported rhodium cata-
lysts are effective [19,20], together with a class of homogeneous
ruthenium catalysts, coupled with halogen promoters and a phosphine
oxide [Zl].


The key element in Texaco's alcohol/ester process is the choice

of ruthenium "melt" catalyst [11,14]. The essential components are
a ruthenium source, typically ruthenium(IV) oxide, ruthenium(III)
acetylacetonate and triruthenium dodecacarbonyl, dispersed in a
molten, quaternary phosphonium or ammonium salt, such as tetrabutyl-
phosphonium bromide and tetrabutylphosphonium iodide. Hydridotri-
ruthenium undecacarbonyl has been implicated in the synthesis gas
conversion to oxygenates, such as ethylene glycol [11], but other
qdded ligand or metal modifiers may be used to alter the final prod-
uct composition. The ruthenium source and quaternary Group VB salt
are generally solids at ambient temperatures, but the melting point
of the salt must lie well below the temperature necessary for CO
hydrogenation (typically ZOO-Z50°C). Under typical operating condi-
tions, then, the quaternary salt provides a highly polar, fluid
medium for solubilization of the ruthenium active catalyst and for
effecting the desired CO hydrogenation to aliphatic oxygenates.
Other important features of ruthenium "melt" catalysis include: a)
the high productivity of these systems--with liquid weight gains
exceeding 100% in batch experiments and productivities of 0.4 g/cc/hr
in continuous processing, b) the broad range of oxygenates that can
be generated, including specific alkanols [22-24], esters [25],
diols [11], and carboxylic acids, and c) the ease by which the prod-
uct alcohol and ester products may be isolated, by fractional distil-
lation of the crude liquid product, and the solid residual ruthenium
catalysts recycled [2Z] to yield the same product compostion at sim-
ilar rates of CO hydrogenation.

Ruthenium Catalyst Choice

Table 1, expts. 1-8, illustrates the general synthesis of

Cl-C3 aliphatic alcohols and their acetate ester derivatives. Meth-
anol, ethanol, and their acetate esters, may each be the predominant
product. The catalyst precursor generally comprises ruthenium(IV)
oxide, hydrate, a tetrabutylphosphonium halide and optionally one
or more cocatalyst species selected from the group including halogen-
free titanium or zirconium compounds [23], a cobalt derivative [22]
and a halogen-free manganese or rhenium compound [24]. The catalysts
are heated under synthesis gas to at least 150°C and 35 atm in order
to achieve the desired CO-hydrogenation. In particular, it may be
seen from an inspection of Table 1 that:

1) Ethanol may constitute up to 59% of the crude liquid prod-

uct, with 76% being Cl-C3 alcohol, when employing the Ru02·Bu4PI
couple (expt. 1). The ruthenium-hal ide-free titanium-Bu4PBr com-
binations [23] likewise produce Cl-C3 alcohols in 61% selectivity
of which up to 66% is ethanol (see expt. 4 and 5).

2) A balanced mix of Cl-C3 alcohols and acetate esters may be

realized in high yields using a variety of ruthenium-cobalt bime-
tallic catalysts [22], as illustrated here by expts. 2 and 3.

3) Methanol will be the predominant product fraction when the

catalyst comprises ruthenium plus a halogen-free rhenium or manganese
compound [24]. With Ru02-Re2(CO)lO/Bu4PBr (expt. 6), methanol selec-
tivity in the liquid product is 85 wt % and the turnover frequency
is 24 moles MeOH/g atom Ru/hr.

4) The addition of certain solvents may lead to further changes

in product composition [25]. The presence of added p-dioxane, for
example, in expt. 8, makes ethyl acetate the predominant product,
with alkyl acetates providing 67% of the organic liquid fraction.

Continuous Processing

Process parameters for the direct generation of Cl-C3 alcohols

and their esters from CO/H2 have been defined using a mechanically-
stirred, high pressure reactor, fitted with continuous gas and
liquid feeds, a gas-liquid product separator and extensive liquid
product traps. For a standard, triruthenium dodecacarbonyl-tetra-
butylphosphonium iodide catalysts precursor, fed to the reactor in
an alkanol diluent, both the product composition and the catalyst
productivity are found to be very sensitive to the choice of the
following operating parameters: a) temperature, b) synthesis gas
composition, c) ruthenium catalyst concentration and d) the gas and
liquid feed rates, as well as e) the nature of the alkanol diluent
used to initially solubilize the RU3(CO)12-Bu4PI. These sometimes

Table 1. Alkanols Plus Acetate Esters from Synthesis Gasa

Liguid Eroduct comEosition (wt %)b

Quaternary Alcohols Acetate esters Liquid
Expt. Catalyst precursor salt m.p. MeOH EtOH PrOH MeOAc EtOAcc PrOAcc yield (%)d

1 Ru02 BU4P1 96 12.9 59.5 3.7 3.3 6.4 0.6 151

2 Ru02-Co(acac)3e BU4PBr 100 10.2 28.2 13.7 5.9 19.9 9.5 215
3 RU3(CO)12-Co2(CO)8e BU4PBr 100 9.3 29.1 12.6 7.0 21.1 11.9 230
4 Ru02-Ti(acac)2(OBu)2 BU4PBr 100 16.3 40.7 4.3 3.5 8.0 4.3 169
5 Ru02-Ti(OMe)4 BU4PBr 100 21.0 39.0 5.4 3.6 6.1 3.1 239
6 Ru02-2Re2(COho BU4PBr 100 84.7 9.4 0.4 0.3 85
7 Ru02-8MnC03 BU4PBr 100 86.3 2.3 1.3 0.5 85
8 Ru02-~C02(CO)8f C7H15Ph3PBr 179 <1 4.0 <1 7.0 54.0 6.0

aTypical operating conditions: 220°C; 430 atm constant pressure; CO/H2 (1:1); catalyst charge:
Ru, 2-8 mmole, quaternary salt, 10-25 g.
bLiquid product analysis by glc, also detected: water, alkyl formates, butanols, acetaldehyde,
2-methyl-l,3-dioxolane, dialkyl ethers, glycol monoalkyl ethers and ethylene glycol; C02 and CH4
present in the product gas samples, along with larger quantities of unreacted CO/H2.
cSmall quantities of methyl propionate and ethyl propionate present in these fractions. to.
dLiquid yield (wt %) calculated basis total catalyst charge. "T1
eOperating pressure, 272 atm. A
fRun in p-dioxane solvent, operating pressure 544 atm. Z

conflicting effects are illustrated by the accompanying Tables Z-3

and Figures 1-3.

Figure 1 and Table Z show how both the operating temperature

and the synthesis gas composition can have a very significant influ-
ence upon the alcohol/ester product composition and the productivity
of the system. Within the operating range ZOO-Z70°C, using 1/1 HZ/CO
synthesis gas, the higher the operating temperature the higher the
productivity of oxygenates. Increasing the temperature also in-
creases the proportion of ethanol in the crude product mix (see
expt. 9) and there is an accompanying decrease in methanol concen-
tration. A hydrogen-rich gas (HZ/CO = Z/l), by contrast, tends to
lower the thermal stability of the ruthenium catalyst (see Figure 1),
and to favor hydrogenation over carbonylation activity, so that
ethanol/methanol ratios become smaller and product liquids richer
in methanol (expt. 10).

The known [11,Z6] water-gas shift activity of the ruthenium

catalysts means that any water formed as a by-product of major Cl-C3
alcohol and ester forming reactions (eq. Z-4) will be rapidly con-
verted in situ to additional hydrogen plus carbon dioxide (eq. 5).
Increasing the gas flow, however, increases the partial pressure of
hydrogen and carbon monoxide in the reactor and this in turn leads
to progressively higher productivities of liquid oxygenates (see
Table 3, expt. 11 and Figure Z). At the same time the high gas flow
preferentially sweeps more volatile components, such as methanol,
from the reactor making the liquid effluent measurably richer in
this lighter fraction (expt. 11, column 4).

CO + ZHZ + CH30H + CZH50H + HZO (3)

CO + ZCH30H + CH3COOCH3 + HZO (4)

CO + HZO ~ HZ + COZ (5)

An increase in liquid feed rates also leads to increased liquid

productivity (expt. lZ), and again this is particularly noticeable
for those more volatile products, like methanol and ethanol, whose
critical temperatures (Table 4) are close to the operating temper-
ature of the reactor.

Another parameter of importance in these syntheses is the choice

of optional solvent. Although tetrabutylphosphonium iodide (m.p.
96°C) is a liquid under normal operating conditions (Table Z), and
RU3(CO)12-Bu4PI readily forms a homogeneous liquid under synthesis
gas pressure that can be pumped directly to the reactor, in many
cases it is more convenient to employ an alcohol or ester diluent.


( ·C)


200 •
10 20 30 40 50 60 70

Figure 1. The effect upon alcohol/ester (liquid oxygenate) produc-

tivity of operating temperature and synthesis gas compo-
sition. Operating conditions as per Table 2.

Various CI-C4 alcohols have been tested, including the typical prod-
ucts of Table 1. To some degree the solvent s.tructure has been found
to influence final product composition. Where, for example, CO
hydrogenation is conducted at temperatures close to, or higher than,
the critical temperature of the solvent alcohol (see Table 4), the
latter will of course be for the most part in the gas phase. In
turn this leads to changes in the molar concentration of ruthenium
"melt" catalyst in the reactor and an alteration in the effective
productivity of the system. The net increases in observed produc-
tivity, illustrated in Figure 2 for syntheses conducted at 240°C are
correlated in eq. 6:

BuOH < EtOH < MeOH (6)

There are additional complications in switching solvents, how-

ever. For example, ethanol usage tends to inhibit methanol homol-
ogation (eq. 3) [27] and thereby increase the methanol/ethanol ratio
in the liquids generated from synthesis gas. Under equilibrium
conditions, however, raising the ruthenium concentration does lead
to significantly improved liquid oxygenate productivity (see Figure
3) with each of these alkanols.

Table 2. The Synthesis of Alcohols and Esters from CO/H2.

Effect of Operating Temperature and Synthesis Gas Composition

UEon Liguid Product ComEositiona

Operating CO/H2 Liguid Eroduct comEosition ~wt %)

Expt. temp (OC) composition MeOH EtOH PrOH MeOAc EtOAc

9b 200 1/1 41.4 26.8 5.2 2.9

220 1/1 38.3 21. 7 5.2 0.4
240 1/1 43.0 34.3 5.3
250 1/1 39.0 39.3 5.2 1.2 0.6
260-270 c 1/1 36.1 35.4 4.2 0.9 0.9
10d 220 1/2 64.0 21.6 1.3
230 1/2 69.5 21.6 0.7 0.2
240 1/2 73.9 20.4 0.3 0.2
250 c 1/2 17 .0 31.7 1.1 1.4

aOperating conditions: gas feed rate, 200 STP l/hr; liquid feed
rate, 20 cc/hr; 476 atm total pressure.
bExperimental series where gas composition is 1/1: liquid feed
contains 7555 ppm ruthenium (as RU3(CO)12)' 36.8% BU4PI, and the
remainder is n-butanol.
CCatalyst decomposition is evident in this temperature range.
dExperimental series where gas composition is 1/2: liquid feed
contains 7318 ppm ruthenium (as RU3(CO)12), 37.1% BU4PI and the
remainder is n-butanol.

Catalyst Life

Long-term, multiple recycle studies have been completed for

the triruthenium dodecacarbonyl-tetrabutylphosphonium iodide couple
and typical data are summarized in Figure 4. No significant chang-
es in catalyst productivity during the course of such experiments
were noted. The only apparent trend is a moderate increase in
ethanol concentration in the crude effluent, coupled with a moder-
ate methanol decrease, that leads to rising ethanol/methanol
ratios, ranging from 0.64 to 0.85, as the experiment progresses.


A flow sheet for generating Cl-C3 alcohols plus their acetate

esters from synthesis gas is outlined in Figure 5. Synthesis gas
and recyclable catalyst solution are fed to a stirred tank reactor
under pressure. The catalyst solution consists of ruthenium spe-
cies (Table 1) solubilized in quaternary phosphonium salt plus
alcohol-ester bottoms. Hydrogenation of CO is conducted at ca.

Table 3. The Synthesis of Alcohols and Esters from CO/H2

Effect of Gas and Liquid Feed Rates Upon Liquid Product Composition

Gas Liquid Liguid Eroduct comEosition (wt %) Productivity

Expt. (STP l/hr) (cc/hr) MeOH EtOH PrOH MeOAc EtOAc (g/hr)

lla 180 20 46.2 33.4 4.3 2.4 3.7 24.6

331 20 48.5 30.9 4.1 2.4 3.6 36.7
421 20 49.6 29.6 3.6 2.1 3.0 40.8
l2 b 450 20 37.9 40.9 3.6 1.8 3.5 35.1
450 60 39.9 48.0 3.3 1.6 3.0 45.4

aExperimental series where liquid feed contains 10,264 ppm ruthenium (as RU3(CO)12)' 47% BU4P1,
and the remainder is methanol; operating conditions are: 240°C; 476 atm pressure; CO/H2: 1/1.
bExperimental series where liquid feed contains 10,264 ppm ruthenium (as RU3(CO)12)' 47% BU4P1
and the remainder is ethanol; operating conditions are: 240°C; 476 atm pressure; CO/H2 : 1/1.

(STP IIhrl



20 30 40


Figure 2. The effect upon alcohol/ester productivity of alkanol

solvent composition and synthesis gas flow rates. Oper-
ating conditions: 240°C; 476 atm; H2/CO: 1/1; liquid feed
rate, 10 cc/hr; initial ruthenium conc., 10,266 ppm.

240°C under pressure (450 atm). After a 3-hour hold period, synthe-
sis gas conversion is ca. 40%; synthesis gas is present in excess.
Productivity is 0.3 grams of liquid oxygenate cc-l hr-l.

Crude reaction product is continually withdrawn from the reactor

to a gas-liquid separator. A gas separation facility is necessary
to vent impurities, notably carbon dioxide and methane. The liquids
from the gas-liquid separator are fed to the first (vacuum) recovery
tower. Here the majority of the Cl-G3 alcohol plus acetate ester
product is taken overhead. Subsequent removal of small quantities
of water from this fraction will be necessary. The 'bottoms' stream
from the vacuum recovery tower is then further fractionated to remove
smaller quantities of higher-boiling liquid oxygenate products, such
as ethylene glycol and glycol ethers. The residual fraction, con-
taining ruthenium-quaternary phosphonium salt, is recycled to the
carbonylation reactor. A slipstream from this catalyst-rich liquid
might be sent to a heavy-ends tower for reprocessing of any deacti-
vated catalyst.


Raw material costs for this Cl-C3 alcohol/ester process have


Table 4. Physical Constants of Solvent Alkanols

Boiling Critical Critical

Solvent point (DC) temperature (DC) pressure (psi)

Methanol 65.0 240 1153

Ethanol 78.5 243 926
n-Butanol 117 290 641

been estimated for two series of ruthenium catalyst compositions

(Ru3(CO)12-Bu4PI and RU3(CO>!2-Co(CO)8-Bu4PBr, see Table 5) as well
as for a broad range of operating conditions. Based on the proc-
essing data summarized in Tables 1-4 and Figures 1-4, these typical
experiments were selected to give some insight into the influence
upon raw material costs of parameters such as solvent composition,
synthesis gas composition, liquid feed rates, gas feed rates and
operating temperature. Also considered were the effects of gener-
ating alkanol-rich products versus an alcohol/ester mix. It is evi-
dent from an inspection of Table 5 that:

1) Raw material costs as low as 74¢/gallon for an alcohol-rich

product can be realized when the process is operated at the condi-
tions of Example II.

2) Carbon efficiency is highest in the case of the alcohol/ester

generation using the ruthenium-cobalt bimetallic catalyst series
(Example VI), although productivity is low.

3) Operating conditions selected so as to achieve maximum proc-

ess productivity, as in Examples II and V, may in some cases also
lead to the lowest raw material costs.


Numerous factors must be considered in the selection of organic

oxygenates as blending components in gasoline, particularly cost,
octane blending value, volatility, water and gasoline compatibility,
raw material availability; engine performance, and corrosion, as
well as environmental problems [28J. In this study, we have focused
on two different Cl-G3 alcohol/ester blends produced by CO hydrogen-
ation using the process outlined above. The first blend (see Table
6, product mix "A"), rich in methanol and ethanol, is similar in
composition to those prepared by the procedures of Table 2, expt. 9.
The second blend ("B"), also contains ca. 52% methanol plus ethanol,
but in addition there is a sizable quantity of methyl and ethyl ace-
tates. This more closely approximates the products of Table 5, Ex-
ample VI, and is generated using the Ru-Co bimetallic catalyst



(ppm.102 )



20 40 50


Figure 3. The effect upon alcohol/ester productivity of ruthenium
catalyst concentration. Operating conditions: 240°C;
476 atm; H2/CO: 2/1; gas flow rate, 250 STP l/hr; liquid
feed rate. 20 cc/hr.

system. Both blends are, of course. the products of numerous cata-

lyst cycles under standardized operating conditions. They have been
isolated. with the minimum of fractionation, by distillation of the
appropriate crude liquid effluents.

The excellent octane blending values of Cl-C3 alcohols are well

established in the open literature [28,29]. but the potential advan-
tage of a higher ester content composition (as in blend B) is ex-
pected to be in its lower gasoline extractability by water. a How-
ever. the particular alcohol/ester composition prepared and tested
here. when blended into gasoline. demonstrated certain other prop-
erties, particularly in the areas of corrosion. gum-forming tenden-
cies and oxidative stability. which were significantly poorer than
those seen with the Cl-C3 alcohol mix A. There is a further concern

aNo significant extraction noted when 10 vol % of either methyl

acetate or ethyl acetate in 100 ml gasoline was heated with 10 ml
water. Methanol at 10 vol % in gasoline under similar treating
conditions is almost completely extracted.



a --a
_ _ _ _- : - a - a - - - - =
ALCOHOL a_a-a a \.
CONTENT (wt '11.1 30



. _9----9-- 9--9-9

100 200 300 400 500


Figure 4. Ruthenium catalyst life study--effect upon product com-
position. Operating conditions: 230°C; 476 atm; H2/CO
1/1; gas flow rate, 100 STP l/hr; liquid feed rate, 20
cc/hr; initial ruthenium conc., 5733 ppm; solvent,

with this type of alcohol/ester composition in gasoline, that the

methyl and ethyl acetates might undergo at least partial hydrolysis
to yield significant quantities of free acid.

Preliminary performance data for our two blends, A and B, may

be summarized, as follows:

1) The octane improvement properties of the Cl-C3 alcohol mix,

A, appear to be good, and on this basis it may show promise as future
high octane blending agent. The ~ON and ~RON values a (ca. 3.4 and
1.4, respectively, for mix A) are noticeably higher than those for
the composite Cl-C3 alcohol-ester mix, B. The increase in volatil-
ity, as measured by ~RVP,b is greater, on the other hand, for the

aOctane results by blending 10 vol % alcohol or alcohol-ester into

medium octane unleaded base fuel (RON 92, MON = 82, and
(R + M) /2 = 87).
bvolatility results by blending 10 vol % alcohol into medium vola-
tility fuel, where RVP of base fuel is 10.0. ASTM D-323 test
was used.




Figure 5. Alcohol/ester generation, flow diagram

alcohol/ester gasoline blend.

2) The oxidative stability and gum-forming tendencies of the

gasoline-containing alcohol/ester mix appear poorer than for gaso-
line containing mix A. Stability of mix A is equivalent to that of

3) The alcohol composition in gasoline exhibits superior corro-

sion properties, when compared to the Cl-C3 alcohol/ester blend, for
each of the metals commonly used in automotive engine surfaces.


1) A process for generating CI-C3 alcohols and their acetate

esters has been described wherein the catalysts comprise a ruthenium
source plus a low-melting quaternary phosphonium salt. The optional
addition of halide-free cobalt, Group IVA and Group VIlA metal salts,
or certain solvents, may further alter the product composition and
make methanol, ethanol and alkyl esters, each the predominant

2) A process schematic and raw material costs have been devel-

oped for the proposed process, based on our understanding of the
effect of critical operating parameters upon the composition of the
CI-C3 alcohol/ester product and the productivity of the ruthenium

3) Octane response and related fuel improvement properties for

two typical product mixtures are discussed.
Table 5. Alcohol/Esters from Synthesis Gas: Raw Materials Costs ~

Example I II V VI
-III- -IV-
Catalyst composition . RU3 (COh2-Bu4PI RU3(CO)12-Co2(CO)8-Bu4PBr
Solvent n-BuOH n-BuOH EtOH MeOH EtOH c

Synthesis gas (CO/H2) 1/2 1/2 1/1 1/1 1/1 1/1

Operating conditions:
Liquid feed rate (g/hr) 20 20 20 20 60 20
Gas feed rate (l/hr) 100 250 450 341 450 100
Temperature (OC) 220 240 240 240 240 220

Cl-C3 alcohols (g/hr) 8.4 39.0 31.0 29.1 36.6 8.7
Cl-C3 alkyl acetates (g/hr) 0.1 1.1 2.8 3.1 3.6 3.8

CO conversion per pass (%) 44 68 26 39 37 43

Carbon selectivity to liquid

product (%)a 75 74 56 61 59 94

Synthesis gas volume (Mscf)

per lb of liquid product ~
0.048 0.043 0.049 0.057 0.058 0.047
Raw material cost (¢) per "z
gallon of liquid product b 82 74 84 97 99
aLess in situ C02 generation. Z
bAssuming cost of synthesis gas is $2.50 Mscf; density of liquid product, 0.82 g/cc. m
cSolvent is Surfonic »

Table 6. Composition of Alcohol/Ester Product Mixtures

Volume (%)
Product mix "A" Product mix "B"
Component C1-C3 alcohols C1-C3 alcohol/esters

Methanol 46.6 30.3

Ethanol 40.4 22.1
iso-Propanol 1.0
n-Propano1 1.4 2.2
n-Butano1 3.5
n-Pentano1 1.9
Methyl acetate 0.7 13.4
Ethyl acetate 2.3 15.5b
Propyl acetate
Higher oxygenates 8.6 a
100.0 100.0
alnc1udes: 1,3-dioxo1ane and its alkyl derivatives, a1koxymethanes,
a1koxyethanes, alkyl formates and alkyl propionates.
blnc1udes some alkyl propionates, substituted dioxo1anes, mono and
dia1koxyethanes, and dia1koxymethanes.


1. Hydrocarbon Process., 61(10):21 (1982).

2. S. S. Stinson, Chem. Eng. News, p 35 (June 25, 1979).
3. B. Tanigushi and R. T. Johnson, Chemtech, 9:502 (1979).
4. European Chem. News, p 14 (May 31, 1982).
5. E. G. Guetens, J. M. DeJovine and G. J. Yogio, Hydrocarbon
Process., 61(5):113 (1982).
6. Chem. Eng. News, p 14 (June 14, 1982); Chem. Week, p 52
(June 24, 1981).
7. G. A. Petzet, Oil Gas J., p 35 (September 6, 1982).
8. J. L. Keller, Hydrocarbon Process., 58(5):127 (1979).
9. Chem. Eng. News, p 21 (April 5, 1982).
10. Chem. Eng. News, p 37 (April 7, 1980).
11. J. F. Knifton, J. Am. Chem. Soc., 103:3959 (1981).
12. J. F. Knifton, Hydrocarbon Process., 60(12):113 (1981).
13. J. F. Knifton, J. Cata1., 76:101 (1982).
14. J. F. Knifton, U.S. Patent 4,265,828 (1981); assigned to
Texaco Development Corp.
15. A. Sugier and E. Freund, U.S. Patent 4,122,110 (1978); assigned
to Institut Francais du Petro1e.
16. A. Sugier and E. Freund, U.S. Patent 4,291,126 (1981); assigned
to Institut Francais du Petro1e.
17. P. C. E11gen and M. M. Bhasin, U.S. Patent 4,014,913 (1977);
assigned to Union Carbide Corp. and related patents.

18. M. M. Bhasin, W. J. Bartley, P. C. E11gen and T. P. Wilson,

J. Cata1., 54:120 (1978).
19. E. I. Leupold, H. J. Schmidt, F. Wunder, H. J. Arpe, H. Hachen-
berg, German Offen. 2,846,148 (1980); assigned to Hoechst
20. M. Ichikawa, J. Chem. Soc., Chem. Comm., 566 (1978); Chemtech,
12:674 (1982).
2l. B. K. Warren, U.S. Patent 4,301, 253 (1981); assigned to Union
Carbide Corp.
22. J. F. Knifton and J. J. Lin, U.S. Patent 4,332,915 (1982);
assigned to Texaco Develop. Corp.
23. J. F. Knifton, U.S. Patent 4,339,545 (1982); assigned to Texaco
Develop. Corp.
24•• J. F. Knifton, U.S. Patent 4,332,914 (1982); assigned to Texaco
Develop. Corp.
25. J. J. Lin and R. G. Duran1eau, U.S. Patent pending.
26. J. F. Knifton, J. Mol. Cata1., 11:91 (1981).
27. J. F. Knifton and J. J. Lin, U.S. Patent to issue.
28. H. Beuther and T. P. Kobylinski, Symposium on Chemistry of Oxy-
genates in Fuels, Paper #45, 184th ACS Meeting, September
12-17, 1982.
29. Oil Gas J., p 62 (June 26, 1978).


Cheryl K. Rofer-DePoorter

Geochemistry Group
Los Alamos National Laboratory
P.O. Box 1663
Los Alamos, NM 87545


The water gas shift reaction is catalyzed by numerous metals

and oxides. Thus, in any environment containing CO and HZO, or COZ
and HZ, the water gas shift reaction,

CO + HZO t COZ + HZ, (1)

or its reverse, the water gas reaction, may occur. The water gas
shift is used to regulate HZ and CO concentrations in synthesis gas,
and it occurs as part of synthesis gas reactions such as the Fischer-
Tropsch synthesis, in which CO and HZ are starting materials, and
COZ and HZO may be products. At typical synthesis gas reaction tem-
peratures, the equilibrium constant for the water gas shift is close
enough to unity and the reaction proceeds rapidly enough that all
four species can be expected to be present. The economics of syn-
thesis gas reactions usually require that oxygen rejection be via HZO
rather than COZ; therefore, control of the oxygen-carrying product
through the water gas shift may be desirable.

Although the water gas equilibrium is sometimes conceptually

separated from other synthesis gas reactions, it probably shares el-
ementary reactions with them and thus cannot be separated either in
a theoretical or practical way. This is an important point; if the
water gas shift itself is considered an elementary step (which it
cannot be in a heterogeneously catalyzed system), its function in a
synthesis gas reaction such as Fischer-Tropsch cannot be understood.
It is tautological to attribute, for example, the production of HZO
as the oxygen-carrying product for an iron-catalyzed Fischer-Tropsch

system to iron's activity in the water gas shift reaction. The dif-
ferences between HZO- and COZ- producing catalysts are rather within
some subset of elementary reactions occurring under Fischer-Tropsch
conditions to give an effective water gas shift. Identification of
these elementary reactions will allow the design of improved Fischer-
Tropsch catalysts.

In the conventional phenomenological approach to catalytic ki-

netics, a mechanism is found that gives orders of reaction similar
to those in the empirical rate equation. The approach I will take
in this paper is complementary to the phenomenological approach, but
it is seldom used [1]. Studies of elementary reactions will be em-
phasized in order to provide a basis for suggesting their participa-
tion in the water gas shift.

Increasing availability of experimental data and of computer

calculational capability makes this approach more useful than it has
been in the past. A calculational analysis of the mechanism built
up from the elementary reactions is not possible within the scope of
this paper, but work is in progress on estimating (or obtaining from
the literature) rate constants for elementary reactions of interest
and combining them to give overall reaction rates.

The organization followed in this paper will be similar to that

in my analysis of the Fischer-Tropsch synthesis [1]. The same nota-
tion will be used. However, because the water gas shift reaction
takes place with metal, metal oxide, and homogeneous catalysts, those
subdivisions will be used. Relevant material will be evidence for
intermediates or reactions on (a) the metals in the catalysts used
industrially, and (b) other metals. Direct observation of interme-
diates or reactions will be preferred to inference from kinetics, and
systems in which elementary reactions are isolated will be preferred
to systems of complex reactions. The literature has been covered
from 1960 through mid-198Z, with emphasis since 1970.


Metal catalysts are of particular interest for understanding the

relation of the water gas shift reaction to the Fischer-Tropsch syn-
thesis, because of the reduced nature of the Fischer-Tropsch cata-
lysts under reaction conditions. The evidence for the relevant el-
ementary reactions has been reviewed [1]. Publications since that
time have generally supported those reactions. Deuterium adsorbs
dissociatively on Rh(lOO) [Z]; potassium on iron surfaces increases
the adsorption energy of hydrogen [3]. Adsorption of hydrogen is
hindered by adsorbed electronegative atoms, such as oxygen [4],
carbon [4,5], and nitrogen [5]. Several techniques have given more
detail on the adsorption of CO on nickel [6], alumina-supported
cobalt [7], platinum [8], copper [8], and rhodium [Z]. The degree

of dissociation of CO on rhodium is subject to differing interpre-

tations [9], and CO is reported to dissociate on stepped nickel
surfaces [10], but not on kinked platinum surfaces [11]. The pres-
ence of adsorbed potassium on platinum strengthens the adsorption of
CO [12], as do sodium, potassium, and cesium on nickel, in addition
to inducing CO dissociation [13]. Conversely, an oxygen adlayer on
Cu-Ni surface alloys weakens CO adsorption [14]. These effects on
CO adsorption are generally interpreted to mean that the alkali
metals increase backbonding from the metal to CO, and electronegative
elements decrease the backbonding.

A thermal desorption study of CO and H20 adsorbed on Ru(OOl)

shows that the presence of water induces a strongly-bound state of
CO, possibly dissociated [15]. Carbon dioxide dissociates on alumi-
na-supported rhodium to CO and adsorbed 0, although the details of
this interaction, and the effect of hydrogen on it, are not clear
[16]. Coadsorption of C02 and H2 on alumina-supported rhodium gave
adsorbed formate, as observed by infrared spectroscopy [17].

The overall kinetics of the water gas shift reaction have been
studied on unsupported iron [18] and platinum [19], on alumina-sup-
ported Group 7B, 8, and lB metals, some of which were also supported
on silica and carbon [20], on zeolite- and alumina-supported rhodium
[21], and on zeolite-supported ruthenium [22].

For unsupported iron at high temperatures, an oxygen transfer


C02 (g) t CO(g) + O(ads) (2)

H20(g) ! H2(g) + O(ads), (3)

was deduced [18]. However, this is in conflict with findings that CO

adsorbs readily on metals and H2 adsorbs dissociatively (see discus-
sions in Ref. 1, and references therein). For unsupported platinum,
the stoichiometric number method [23] was used to interpret overall
kinetics and isotopic labeling experiments. The proposed mechanism
[19] is given in Table 1. The adsorbed intermediate was not iden-

For supported metals, the situation appears to be more complex.

Because the commonly-used supports, such as alumina, can adsorb water
to a significant extent and the other components of the water gas
shift equilibrium to lesser degrees, the catalyst may be bifunction-
al, with some processes taking place on the metal surfaces, some on
the support surfaces, and some at the interface between the metal and
the support. A comparative study [20] of the water gas shift reac-
tion on rhenium, cobalt, iron, nickel, rhodium, ruthenium, palladium,
iridium, osmium, platinum, copper, and gold showed activities rang-
ing over three orders of magnitude for the metals supported on alu-
Table 1. Mechanism Proposed by Masuda and Miyahara [19] for the
Water Gas Shift Reaction on Unsupported Platinum

co + M ! M-CO
H20 + 2 M t M-H + M-OH
M-CO + M-OH ! adsorbed intermediate
adsorbed intermediate t M-C02 + M-H
M-C02 t M + C02
2 M-H t 2 M + H2

mina. Platinum and rhodium were found to be most active on alumina

supports, of intermediate activity on silica, and least active on
carbon. A mechanism was proposed for a bifunctional catalyst. This
mechanism is given in Table 2, where M indicates a site on the metal
and MO a site on the support. However, the activity of a bifunction-
al catalyst should be strongly dependent on the dispersion of the
metal, and no such dependence was found.

Methane was a significant side-product in the water gas shift

reaction over zeolite- and alumina-supported rhodium, and both meth-
ane and ethane were observed in small amounts over the alumina-sup-
ported catalyst [21]. The activity of alumina alone for the reaction
was two orders of magnitude less than for the alumina-supported rho-
dium. Zeolite-supported ruthenium also produced methane in addition
to the water gas shift reaction, but the support seems to have less
effect than for rhodium [22]. The occurrence of methanation greatly
complicates the interpretation of these results.

A set of elementary reactions describing the water gas shift

reaction on most metals is given in Table 3. This table includes the
reactions of Table X in Ref. 1, with all adsorption-desorption steps
made explicit, and nucleophilic attack of adsorbed OH on adsorbed CO
added, partly because of more recent studies [17,20], and partly be-
cause of evidence on the homogeneously catalyzed reaction, to be dis-
cussed later in this paper. Hydroxycarbonyl is given as the product
of nucleophilic attack, rather than formate, because several reac-
tions can be written for formate formation and decomposition, none
of which have significant support from surface studies. Although in-
clusion of formate at this time would be both speculative and
lengthy, it may be justified with further study.


The water gas shift catalysts used industrially have been iron
and chromium oxides (high-temperature catalyst) and copper, zinc,
chromium, and aluminum oxides (low-temperature catalyst). Limited

Table 2. Mechanism Proposed by Grenoble,

Estadt, and Ollis [20]

H20 + MO ~ HO-MO-H
H20 + MO ~ H20-MO
H20 + 2 M ! M-OH + M-H
CO + M ! M-CO
M-CO + H20-MO t OM-O(H)C(H)O
OM-O(H)C(H)O + M ! M-O(H)C(H)O
M-O(H)C(H)O t M + C02 + H2

information is available on the adsorption of H2' H20, CO, and C02

on these oxides. However, many of the elementary steps of the water
gas shift are adsorption-related: adsorption and dissociation, or
their reverse, association and desorption.

Hydrogen Adsorption

At least three, and as many as seven, types of hydrogen have

been identified as adsorbed species on zinc oxide [24-32]. Infrared
spectroscopy has identified three species: a hetero1ytica11y disso-
ciated species, Zn-H and OH on adjacent sites, that appears to be
active in hydrogenations [24-27]; a low-temperature molecularly ad-
sorbed species that participates in H2-D2 exchange [28,29]; and a
species inactive for hydrogenation that appears to be bridged, Zn-H-
Zn or O-H-O [26]. Temperature-programmed desorption (TPD) studies
and conductivity studies seem to indicate the presence of greater
numbers of hydrogen species [30-32], but these have not all been
identified with infrared species. At least two TPD peaks appear to
represent different reaction paths to desorption for the species
identified in the infrared studies [33].

Although zinc oxide is considered the hydrogen activation compo-

nent in methanol and water gas shift catalysts, other components of
the catalysts can also adsorb hydrogen. TPD results [34] show five
different types of hydrogen adsorbed on gamma-alumina between -196
and 450°C. Chromia and chromia-si1ica adsorb hydrogen, although to a
lesser extent than zinc oxide. A very low-temperature form of hy-
drogen adsorbed on chromia appears to be molecular, analogous to the
low-temperature form on zinc oxide [35]. A heterolytic dissociative
adsorption of hydrogen is reported at intermediate temperatures, and
transformation of the Cr 3+ bonded to H to Cr 2+ and OH is reported to
take place at higher temperatures [36].

The reactions producing the various forms of hydrogen adsorbed

on oxides are likely to be the following:

Table 3. Proposed Mechanism of Water

Gas Shift on Metals

HZ + M t M-HZ(physisorbed)
M-HZ(physisorbed) t H-M-H
CO + M t M-CO(physisorbed)
M-CO(physisorbed) t M-CO
HZO + M t M-HZO(physisorbed)
M-HZO(physisorbed) t M-OHZ
COZ + M t M-COZ(physisorbed)
M-COZ(physisorbed) t M-COZ
M-CO t M-~I
M-1IIC ++ O-M-C
M-OH t H-M-O
M-CO + M-O M-COZ +
M-CO + M-OH t M-C(O)OH
M-C(O)OH +

HZ + MO t MO-HZ(physisorbed) (4)

MO-HZ(physisorbed) t H-MO-H (5)

H-MO-H + Z MO t OM-H-MO-H-OM (6)

Other routes are possible, and diffusion on the surface may be

important, as indicated by the TPD results. Reactions 4-6 account
in a simple way for relationships among the three species observed
by infrared spectroscopy, but the reactions themselves have not been
observed directly.

Water Adsorption

The adsorption of water onto metal oxides gives hydroxyl groups,

intermediates also formed by the adsorption of hydrogen. In the case
of water, dissociation is to a hydrogen that adds to an oxide oxygen
and a hydroxyl that adds to a metal ion. Hydrogen bonding with the
oxide oxygens and hydroxyl groups also causes molecular adsorption
of water.

On iron oxide, adsorption isotherms [37], dielectric relaxation

[38], infrared spectroscopy [39], and ultraviolet photoemission spec-
troscopy [40] have shown both molecularly absorbed and dissociated
water. Water is physisorbed onto surface hydroxyls. Physisorbed

water was observed by infrared spectroscopy on an iron-chromium cat-

alyst; dissociative chemisorption was also deduced [41].

Water adsorption on zinc oxide has been studied by adsorption

isotherms [42] and a combination of thermogravimetry and infrared
spectroscopy [43]; again, molecular adsorption and dissociation were
both observed. Water adsorbs dissociatively onto oxygen-treated cop-
per and zinc [44], although it adsorbs poorly onto clean copper and
zinc surfaces. For a low-temperature shift catalyst, both hydroxyls
and undissociated water have been observed by infrared spectroscopy
[45]. Dissociation increases with the degree of reduction of the
catalyst. Molecularly adsorbed water appears to be present on both
catalyst and support.

On alumina, as determined by microcalorimetry and infrared spec-

troscopy [46], water is physisorbed, then molecularly chemisorbed
and dissociated. Silica [47] and chromia [48] behave similarly. The
processes of water adsorption on silica-alumina appear to be similar
to those on silica and alumina, but the combination has a greater
capacity for water adsorption than what would be extrapolated by a
linear combination of the two [49].

The reactions of water adsorption on oxides can be summarized:

H20 + MO t H20-MO(physisorbed) (7)

H20- MO (physisorbed) t H20- MO (8)

H20- MO (physisorbed) t MO-HOH (9)

H20- MO t HO-MO-H (10)

MO-HOH t HO-MO-H (11)

There appear to be two kinds of water chemisorbed molecularly: one

in which the oxygen is bonded to the metal ion (reaction 8), and one
in which the water is hydrogen bonded to the oxygen or hydroxyl
groups of the oxide (reaction 9).

Carbon Monoxide Adsorption

CO adsorption on oxides has been less studied and therefore is

not as well understood as CO adsorption on metals. Infrared spec-
troscopy has been the method of choice for examining CO adsorption
on metal oxides; few other methods have been used. In general, a
low-frequency group of bands (1100 to 1800 cm- l ) is attributed to
carbonate, bicarbonate, formate, and carboxyl or hydroxycarbonyl, and
a higher frequency group of bands (2000 to 2250 cm-l ) is attributed
to a weakly adsorbed species, in which the carbon is sigma-bonded to
the metal ion with no backbonding or the interaction is primarily

electrostatic. Carbonyl bands (1900 to 2000 cm-l ) are sometimes ob-

served. Interpretation of some of the bands is not unambiguous.
Harrison and Thornton [50] have concluded that the weakly bonded CO
is adsorbed, carbon down, perpendicular to the surface, as in carbon-
yls, but the bonding is primarily electrostatic. Copper is an excep-
tion to this, in that significant pi bonding appears to be present,
and this model is not completely consistent with other data, sug-
gesting the possibility of some pi bonding for other metals. Infra-
red adsorption frequencies for CO adsorbed on transition-metal ions
on silica [51] appear to be consistent with this model, and a simple
correlation between heat of adsorption and CO frequencies for several
oxides [52] also argues for a simple bonding model. The c-o bond
for this type of CO adsorption is strengthened relative to its gas-
phase value.

Conductivity data suggest a partly cationic CO adsorbed on the

Fe 2+ ions of Fe203 [53], although the charge of the CO changes from
partially positive to partially negative with increasing reduction
of the oxide [54]. Carbon monoxide was adsorbed onto a reduced iron-
chromium catalyst with apparent formation of carbonates and carbon-
yls [41]; however, one of the bands attributed to a carbonyl (2095
cm-l ) is in the range usually considered to belong to a weakly-bound
species. On FeO, CO appears to be adsorbed at a metal ion, although
the ultraviolet photoelectron spectrum cannot be fully interpreted
[55]. With potassium on FeO, the CO is adsorbed more strongly, prob-
ably as a carbonyl [55].

Adsorption on copper oxide gives a weakly bound species [56,57],

carbonyls [56-62], and carbonates, bicarbonates, and formates [56,
58,59]; in addition, if another component, such as alumina, silica,
or other metal oxide, is present, bicarbonate and formate concentra-
tions will be increased. A large range of frequencies is observed
for carbonyls (2000-2200 cm- l ); apparently differences in the prepa-
ration of the catalyst affect the relative amounts of copper in the
three oxidation states. It has not been possible to relate frequen-
cies unambiguously to oxidation states because of the lack of appro-
priate model compounds. Copper carbonyls that may simulate surface
compounds have been synthesized only recently [63].

Zinc oxide shows bands in the weakly bound region [61,64-66],

and the carbonate region [64-66], but none that have been identified
as carbonyls. Other surface diagnostics tend to confirm these iden-
tifications [67]. Angle-resolved photoelectron spectroscopy shows
the CO to be adsorbed, carbon end down, on the zinc atoms [68].

Both pure and supported chromia show bands indicating a weakly

adsorbed species [48,69-73], a carbonyl [71], and carbonates [69,72,
73]. Transient response techniques have given results consistent
with these interpretations [74]. However, for a copper chromite
catalyst, it was concluded that carbonyls formed only on copper [56].

Alumina is reported to form one [69,75] or two [76] weakly ad-

sorbed species and carbonates. Kinetic data support the presence of
two types of adsorption, but these cannot be correlated readily with
those identified by infrared spectroscopy [77]. A potassium promoter
on alumina results in carboxylate and carbonate upon adsorption of
CO and C02 [78].

Adsorption of CO on silica is not observed at temperatures

above 75°C [58].

Magnesia forms a weakly adsorbed species and carbonates [59,79],

and there is some evidence for a species containing an unpaired elec-
tron. This species has been identified as a carbonyl radical [80]
or a polymeric radical anion [81]. These species participate in the
disproportionation of CO, forming carbonates and carbon. A similar
species is formed on calcium and strontium oxides [82].

The probable reactions representing CO adsorption on metal

oxides are

CO + MO t OM-CO(physisorbed) (12)

OM-CO(physisorbed) tOM-CO (13)


Distinctions will not be made here between unidentate and bidentate

species. The formation of bicarbonates and formates from CO must
involve interactions with hydrogen or hydroxide and will be discussed
later. The stoichiometries and structures of the polymeric radical
anions formed on alkaline earth oxides are not known, so reactions
cannot be written for their formation and eventual disproportionation
to carbonates and carbon. Therefore, these reactions will be neg-
lected at this time, but they should be considered for catalysts
containing the alkaline earth metals.

Carbon Dioxide Adsorption

Carbon dioxide initially physisorbs on metal oxide surfaces and

then chemisorbs in three ways: with the carbon bonded to metal to
give a carboxyl group, with an oxygen bonded to metal, or with the
carbon bonded to oxygen to give a carbonate. Relatively little evi-
dence exists, however, for the M-OCO species.

Fe203 adsorbs C02 to give a weakly adsorbed species and possible

carbonates, bicarbonates, and carboxyls [83,84]. Chromia and chro-
mia-silica give species similar to those on Fe203 [69,85].

Copper oxide adsorbs C02 weakly [58], with possible formation

of carbonates and carboxyls [86]. Zinc oxide forms carbonates and a

carboxyl upon COZ adsorption [64,83,87]. The carboxyl group proba-

bly does not carry a full negative charge [67].

Two weakly adsorbed species have been identified on silica by

infrared spectroscopy [88], but this observation is in contradiction
to the observation of carboxyl groups on silica exposed to COZ by
electron spin resonance [89] and adsorption-desorption techniques

On alumina, carbonates and bicarbonates [69,91-96] and one [91],

two [9Z], or three [93] weakly bonded species have been observed.
No carboxyls have been reported. An adsorption isotherm study gen-
erally supports the carbonate identifications [97]. The species and
their relative concentrations are strongly influenced by heat treat-
ment and the crystal structure of the alumina [94].

On magnesia, carbonates are formed [79b, 9Zb, 95], and a weakly-

bound species has been observed [79b].

Soluble COZ complexes of cobalt and alkali metal cations have

been characterized [98]. The carbon is bonded to the cobalt, and the
oxygens to the alkali metal cations. The infrared C-O frequencies
are in the range of those observed for carboxyls on surfaces.

The reactions for adsorption of COZ on metal oxides can be


cOz + MO t MO-COZ(physisorbed) (15)

MO-COZ(physisorbed) tOM-C02 (16)

MO-C02(physisorbed) t MO-C02 (17)

MO-C02(physisorbed) t OM-OCO. (18)

Interactions Among Adsorbed Species

The major interactions appear to take place among adsorbed spe-

cies, rather than between adsorbed and gas-phase species. However,
information about these steps is incomplete and sometimes contradic-

The presence of hydroxyl groups on the surface of several of

these oxides appears to promote COZ adsorption, although this effect
has not been studied in detail. Water interacts with adsorbed COZ
on magnesia to produce bicarbonate [99]. This could come about by
nucleophilic attack on the carbon by hydroxyl,

OM-OCO + OM-OH t OM-OC(O)OH + MO, (19)


hydrogen transfer,

MO-C02 + OM-OH t MO-C(O)OH + OM-O (20)

or oxygen insertion,

OM-C02 + OM-OH t OM-OC(O)OH + MO. (21)

In a relevant solution reaction, Cu 2+ has been found to catalyze the

nucleophilic attack of water at a carbonyl carbon [100].

Carbon monoxide and water, as well as C02 and H2, have been co-
adsorbed on zinc and magnesium oxides and the surface intermediates
observed by infrared spectroscopy [101]. Formate ions were the pre-
dominant species observed. Formate [102,103], acetyl and acetate
[102], and carbonate and bicarbonate [103] intermediates have been
trapped during CO-H2 and C02-H2 reactions on copper-zinc catalysts.
Water and CO adsorbed on an iron-chromium catalyst interact to pro-
duce formate [41]. A possible reaction producing formate is

OM-CO + OM-OH t OM-OC(H)O. (22)

This is an oxygen insertion into the M-C bond (or carbon migration
to oxygen) with hydrogen transfer. It may be broken down into

OM-CO + OM-OH t OM-C(O)OH + MO (23)

OM-C(O)OH t OM-OCOH (24)

OM-OCOH t OM-OC(H)O. (25)

Hydrogen addition to adsorbed C02 will also give formate:

M-OCO + M-H t M-OC(H)O + M. (26)

Both lattice and adsorbed oxygen may oxidize CO to C02 on copper

oxide [104,105] and Fe203 [53]. On chromia, adsorbed CO and oxygen
react to give carbonates and carboxy1s [106,107]. The weakly bonded
CO with an infrared absorption at 2187 cm-1 appears to be the reac-
tive species [106]. Coadsorption of CO with oxygen on hydrated mag-
nesia produces infrared absorptions in the carbonate region and no
absorptions characteristic of adsorbed CO [79b]. This reaction may

OM-CO + OM-D tOM-C02 + MO (27)

OM-C02 + OM-O t OM-OC02 + MO. (28)

The interaction between adsorbed CO and hydrogen on zinc oxide

has been investigated by several groups [108-113]. The evidence is

consistent with inductive strengthening of the bonds between the ad-

sorbed species and the zinc oxide, with no new chemical bonds formed
[113]. These species may be more reactive than the individually ad-
sorbed species; otherwise, this interaction is more relevant to
methanation and methanol formation.

Overall Reaction Studies

Studies of the overall water gas shift reaction over iron,

copper, and cobalt-molybdenum catalysts have been reviewed recently
[114]. Empirical rate equations are summarized there and will not
be considered here, although some conclusions on elementary steps in
the mechanism will be reconsidered in the light of the independent
evidence for the elementary reactions. The cobalt-molybdenum cata-
lysts will not be considered here because less information exists for
elementary reactions on those catalysts, and their sulfidation may
introduce significant differences into the mechanism of reaction.

For the iron-chromium catalysts, two mechanisms have been pro-

posed. The regenerative mechanism [115] involves oxygen transfer
through adsorbed oxygen atoms; this essentially consists of reactions
2 and 3 or may also include the adsorption of the other reactants.
Most of the evidence for this mechanism has been obtained at higher
temperatures than are of industrial interest for the water gas shift
reaction, and it may be primarily of interest in metallurgy [18,116,
117]. Mechanisms proceeding through a C-H-O intermediate, such as
the formate ion, have also been proposed [118].

For the copper-zinc catalysts, most of the overall studies [119,

120] conclude that an intermediate, probably a formate [120], is im-
portant in the mechanism. The single dissenting opinion [121] is
based mainly on theoretical considerations and fails to explain sev-
eral points. A comparison of formate decomposition on zinc and mag-
nesium oxides with the water gas shift reaction led to the conclusion
that formate is an important intermediate [lOlJ.

A combined kinetic and infrared study of the water gas shift

reaction on alumina [122] showed that formate ion exists on the work-
ing catalyst and that some formate ions may not participate in the
overall reaction. Infrared spectroscopy suggests that carbonate is
not an important intermediate. Little if any carboxyl was observed.

Although the evidence for most of the steps is poor in terms of

observed intermediates or elementary reactions, a mechanism for the
water gas shift on oxide catalysts is summarized in Table 4.


It is not yet clear to what degree reactions observed in homog-


Table 4. Proposed Mechanism for the Water Gas Shift

Reaction on Metal Oxides

Reaction Number

H2 + MO t MO-H2(physisorbed) (4)
MO-H2(physisorbed) t H-MO-H (5)
H-MO-H + 2 MO t OM-H-MO-H-OM (6)
H20 + MO t H20-MO(physisorbed) (7)
H20-MO(physisorbed) t H20- MO (8)
H20-MO(physisorbed) t MO-HOH (9)
H20-MO t HO-MO-H (10)
MO-HOH t HO-MO-H (11)
CO + MO t OM-CO(physisorbed) (12)
OM-CO(physisorbed) tOM-CO (13)
CO + MOMO t M2C03 (14)
C02 + MO t MO-C02(physisorbed) (15)
MO-C02(physisorbed) tOM-C02 (16)
MO-C02(physisorbed) t MO-C02 (17)
MO-C02(physisorbed) t OM-OCO (18)
OM-OCO + OM-OH t OM-OC(O)OH + MO (19)
MO-C02 + OM-OH t MO-C(O)OH + OM-O (20)
OM-C02 + OM-OH t OM-OC(O)OH + MO (21)
OM-CO + OM-OH t OM-C(O)OH + MO (23)
OM-C(O)H t OM-OCOH (24)
OM-DCOH t OM-OC(H)O (25)
M-OCO + M-H t M-OC(H)O + M (26)
OM-CO + OM-O tOM-C02 + MO (27)
OM-C02 + OM-O t OM-OC02 + MO (28)

eneous metal complexes can be assumed to occur in heterogeneous sys-

tems. However, mechanisms for the water gas shift reaction have been
worked out for several homogeneous systems, and some of the elemen-
tary steps appear to be similar to those in heterogeneous systems.

Mononuclear and polynuclear carbony1s of several transition

metals catalyze the water gas shift. In acidic solution, catalysis
by rhodium [123], ruthenium [124], and palladium [125] complexes has
been observed, and in basic solution, by iron [126-130], nickel
[131], ruthenium [128,129,131-133], rhodium [134-136], osmium [137],
iridium [128,137], platinum [138,139], and chromium, molybdenum, and
tungsten complexes [126,132].

Notation can obscure the similarities between the homogeneous

and heterogeneous reactions. Therefore, in this discussion, reac-
tions will be written in the notation used for the metal surfaces.
Ligands that do not participate in the reaction will be omitted.
Table 5 summarizes the elementary reactions suggested for the homog-
eneous water gas shift reaction. Reactions for both the acidic and
basic media are included.

In proposing a general mechanism for the homogeneous water gas

shift reaction, the charges on metal complexes will not be consid-
ered. Many of the reactions take place with both neutral and charged
complexes (Table 5), and this simplification makes clearer the rela-
tionship to surface reactions. However, the charges will affect the
energetics of the reactions and must be taken into account in calcu-
lations of specific systems.

Nucleophilic attack of hydroxide or water at a carbonyl carbon

has been well documented [140]:

M-CO + OH t M-C(O)OH (29)

M-CO + H20 + H-M-C(O)OH. (30)

The metallocarboxylic acids can break down to give C02 [141]:

M-C(O)OH t M-H + C02. (31)

Carbon dioxide can also react with metal complexes to give formates

C02 + M-H + M-OC(H)O. (32)

The equilibrium
. (33)

also gives formate, which can complex with a metal:

M + HC02- t M-OC(H)O-. (34)

Molybdenum hexacarbonyl reacts with formate ion to give a formato-
molybdate complex and a carboxylate complex [143]. Photochemical en-
hancement of formate decomposition by group 6B metal hexacarbonyls
[144] appears to support dissociation of CO from a carbonyl, followed
by formate complexation and decomposition, as steps in the reaction.
Both monodentate formate [145] and metallocarboxylic acids [141] may
by intermediates, yielding C02 and H2 upon decomposition, but the
metallocarboxylic acid pathway may be energetically more favorable,
at least in the case of iron [145].

Dissociation of H20 and formation of H2,

H20 + M t H-M-OH (35)


Table 5. Previously Proposed Elementary Reactions for Homogeneous

Water Gas Shift Reaction

Reaction Reference

M-CO + OH- t M-C(O)OH- 126,128,130,131

M-CO + OH- t M-H + C02 137
M-CO + OH- t M-H- + C02 133
[M-CO]OH t M-C(O)OH 138
M-CO + H20 t H-M-C(O)OH 128,131
M-CO + Me3N t M-C(0)NMe3 132
M-C(0)NMe3 + H20 t M-H- + C02 + Me3Ng+ 132
M-CO+ + H20 t H-M-C(O)Og+ 129
M-C(O)OH t M-H + C02 128,131,138
H-M-C(O)OH ! M + HC(O)OH 138
M-C(O)OH- + H20 t H-M-C(O)OH + OH- 128,131
M-C(O)OH- t M-H- + C02 126
M-C(O)OH- + B t M-C(0)02- + BH+ 130
M-C(0)02- + H20 t M-H- + HC01- 130
M-(C1)C(0)OH+ +
~ + C02 + HT + C1- 129
M-OC(H)O t M-H + C02 138
M-OC(H)O- t M-H- + C02 124,126,132
H-M-H t M + H2 124,126,128,131,
H-M-H + CO t M-CO + H2 128,131,136
M-H- + CO t M-CO + H- 126,133
M-H- + H20 t M + H2 + OH- 133
M + CO t M-CO 125,126,128,131,
w + CO t M-CO+ 129
~ + CO t M-CO- 133
M + H20 + H-M-OH 136,138
M- + H20 t M-H + OH- 125,126,132,137
M + HC(O)OH +H-M-OC(H)O 138
M + HCOO- t M-OC(O)H- 125,126,132
H-M-OH + S t [H-M-S]OH 136,138
[H-M-S]OH + CO +
[H-M-CO]OH + S 136,138
CO + OH- t HC02- 126,128,131,132
C02 + OH- t HC03- 128,131
HC02- + H20 t H2 + C02 + OH- 128,131
H- + H20 t H2 + OH- 126,133

H-M-H t M + H2, (36)

are known. Addition of water to rhodium [146] and palladium [147]


Table 6. Proposed Mechanism for the Homo-

geneous Water Gas Shift Reaction

Reaction ~umber

M-CO + OH t M-C(O)OH (29)

M-CO + H20 t H-M-C(O)OH (30)
M-C(O)OH t M-H + C02 (31)
C02 + M-H t M-OC(H)O (32)
CO + OH- t HC02- (33)
M + HC02- t M-OC(H)O- (34)
H20 + M t H-M-OH (35)
H-M-H t M + H2 (36)
CO + M t M-CO (37)
H-M-H + CO ~ M-CO + H2 (38)
M-H + CO ~ M-CO + H (39)

complexes followed by elimination of hydrogen has been observed.

Another relevant reaction is the photochemical substitution of water
for CO in Re2(CO)10; the substitution proceeds through a somewhat in-
direct mechanism, however, with photochemical homolysis of the Re-Re
bond as the first step [148]. Under further irradiation, the water
ligand dissociates to give HRe(CO)5 and Re4(CO)12(OH)4.

The addition of CO to an unsaturated metal complex or its sub-

stitution for hydrogen are both likely:

co + M t M-CO (37)

H-M-H + co t M-CO + H2 (38)

M-H + co ~ M-CO + H. (39)

The water gas shift reaction has also been photocatalyzed with a ru-
thenium complex, apparently because of photo-initiation by CO disso-
ciation from the complex [149].

The bicarbonate equilibrium should also be included:


Carbon dioxide can be disproportionated to CO and carbonate by tran-

sition metal dianions [150], in a reaction that may be a part of some
of the homogeneously catalyzed water gas shift and suggests some as-
pects of the heterogeneous reaction. _ Table 6 gives a proposed mech-

A zeolite catalyst prepared from ruthenium hexamine appears to

operate through a similar mechanism. Ruthenium hexamine exchanged
into zeolites shows activity for the water gas shift from 90°C to
230°C [151]. Intermediates identified by infrared spectroscopy sug-
gest a mechanism involving nucleophilic attack by water on CO coor-
dinated to ruthenium to give a ruthenium carboxylic acid.


It should be emphasized for the following discussions that the

evidence for most of the elementary reactions is incomplete. The
reactions on metal surfaces, particularly C02 and H20 association-
dissociation, are the best supported. Many fewer studies are avail-
able for adsorption on oxides, owing to the difficulties of preparing
well-characterized oxide surfaces. The adsorption-desorption reac-
tions are fairly well established, but questions remain as to the
nature of the species formed by CO and C02 adsorption on oxides. The
steps in which the intermediate is formed and decomposed are the
least supported for all mechanisms. It is not clear to what degree
formate, hydroxycarbony1, or oxygen transfer contribute to the over-
all reaction, although most of the evidence points to the participa-
tion of an H-C-O intermediate for the lower-temperature processes.

Considerable similarity exists among the mechanisms proposed

here and previously for the water gas shift on metals, metal oxides,
and in homogeneous solutions. Adsorption and desorption steps on
the surfaces are similar, with the exception of physisorption, to
the association reactions in solution. In all three mechanisms, H2
dissociates to atoms, and H20 dissociates to hydrogen and hydroxyl.
Although the reactions may be similar for CO and C02 adsorption, the
adsorbed species on metals and oxides may have significantly differ-
ent reactivities from each other and from comp1exed CO and C02 in
solution. For CO on metals, backbonding weakens the C-O bond; for
most oxides, backbonding is absent or almost so, and the c-o bond is
strengthened relative to the gas-phase molecule. However, products
of apparent nucleophilic attack on carbon are observed in all media.
Oxidation of CO by adsorbed oxygen atoms on metals is the reverse of
C02 dissociation. This reaction also appears likely on oxides, but
less so in solution. However, it has been observed in one homoge-
neous system [152].

Reactions forming a C-H-O intermediate have not been observed

directly on surfaces. Formic acid or a formate ion adsorbed as
M-02CH is an attractive intermediate because of the stoichiometry of
formic acid and its known surface reactions:

HCOOH t H2 + C02 (41)

HCOOH ~ H20 + CO, (42)


However, the formation of adsorbed formate from CO adsorbed carbon

end down requires several elementary steps of reaction with hydroxyl
and rearrangement (reactions 23-25), or reaction with oxygen and hy-
drogen atoms. None of these reactions have been observed. Formation
from C02 would involve only a hydrogen transfer (reaction 26).
Again, this has not been observed in heterogeneous systems. A large
volume of literature, which cannot be reviewed within the scope of
this paper, deals with the decomposition of formic acid on metal and
metal oxide surfaces. A survey of some of the more recent of this
literature shows great differences, apparently depending on the
method of observation as well as the substrate. For example, most
static methods show that the hydroxyl hydrogen is abstracted by
metals and oxides to give surface formates [153], but a dynamic study
shows that a hydroxycarbonyl radical is formed by the abstraction of
the carbon-bonded hydrogen by the metal surface and that the radical
can detach from the surface [154]. Formation of an anhydride inter-
mediate has been proposed [155], and an anhydride has been observed
by infrared reflectance spectroscopy [156]. However, the lack of
such an observation in the H2-H20-CO-C02 systems suggests that this
is an unlikely pathway for the water gas shift reaction. A beta-
hydride elimination has been suggested as the decomposition route
from a fotmidorhodium complex to C02 and the hydrido complex [157].
Formic acid has also been suggested as an intermediate in the water
gas shift on the basis of its overall kinetics of decomposition on
a low-temperature shift catalyst [120]. However, supporting evidence
in the form of observed elementary reactions would be highly desir-

A simpler set of elementary reactions (reactions 29-31) can be

written for the CO-hydroxycarbonyl-C02 transformation, but few hy-
droxycarbonyl or carboxyl intermediates have been observed on sur-
faces. The observation that iron hydroxycarbonyl complexes decompose
more rapidly than formido complexes [145] suggests that they may be
the kinetically more important intermediates; their short lifetimes
and consequently low steady-state concentrations may prevent their
observation. On the other hand, they may be too unstable to form on

Some of the products appear not to play a part in the water gas
shift reaction except to remove a small amount of material from the
system. Carbonates and the inactive form of hydrogen on oxides are
in this category.


The water gas shift reaction is simpler than the Fischer-Tropsch

synthesis. It commonly achieves equilibrium, whereas Fischer-Tropsch
is at least partly controlled by kinetics. It can be represented by
a conventional balanced chemical reaction, whereas Fischer-Tropsch

must be represented by a generalized chemical reaction, or a collec-

tion of chemical reactions. All this points to the probability that
the water gas shift can be represented by a smaller set of elementary
reactions than Fischer-Tropsch. A set of elementary reactions has
been proposed for the mechanism of the Fischer-Tropsch synthesis [1],
and sets of elementary reactions have been proposed here for the
mechanisms of the water gas shift reaction on metals, on oxides, and
in homogeneous solution. In this section, the relationships between
these two groups of elementary reactions will be examined.

Publications since the proposal of the mechanism for the Fisch-

er-Tropsch synthesis give no reason to alter that set of elementary
reactions. Adsorption of H2 and CO on transition metals, both pure
and modified with promoters and poisons, has been reviewed [158]. A
statistical analysis to evaluate correlations between H2 and CO dis-
sociation and other properties of metals has also been carried out
[159]. Hydrogen dissociation was found to correlate with heat of
vaporization, heat of fusion, molar heat capacity, Debye temperature,
atomic volume, electronegativity, first and second ionization ener-
gies, and electrical conductance. Carbon monoxide dissociation was
found to correlate with heat of fusion, molar heat capacity, elec-
trical conductance, and electronegativity. Several of these param-
eters are related to the metal-metal bond energy, and the others may
relate to electronic effects.

Although attempts to observe formyl intermediates on rhodium

surfaces failed [160], increasing numbers of formyl metal complexes
are being synthesized [161]. The protonation route for hydrogenation
has been further supported [162]. Several methods of C-C bond for-
mation have been demonstrated in complexes [163]. The reverse of
C-H bond formation, the oxidative addition of hydrocarbons to iridium
complexes, has been demonstrated [164]. Gas-phase studies of inter-
actions between metal atoms and hydrocarbons [165] should give in-
formation such as bond energies that can be used in estimating rate
constants for these reactions, and are showing some differences be-
tween the interactions of different metal atoms with hydrocarbons.
The acidity of metal hydrides [166] is another set of information
that will be useful in evaluating mechanisms, particularly in solu-
tion, and possibly in extrapolation to surfaces. Reactions of in-
terest are being studied in early transition metal complexes [167]
and actinide complexes [168], but these reactions are stoichiometric
rather than catalytic. Relevant organometallic reactions have been
reviewed [169].

Several kinetic studies support the participation of carbon

atoms and partially hydrogenated carbon as intermediates [170]. From
observations of this type, it has been suggested that hydrogenolysis
can be regarded as a reverse Fischer-Tropsch synthesis [171]. Oxi-
dation and carburization of iron have been further studied, and it
appears that some iron oxide may be present under synthesis condi-

tions [172]. Changes in kinetic regimes were observed in methanation

over nickel foils [173] and in the production of hydrocarbons and
oxygenates over LaRh03 [174]. Where Schulz-Flory analysis has been
applied to product distributions, both consistent [175,176] and in-
consistent [170a,173,176,177] distributions have been observed.
Product distributions have been modelled for situations other than
the simple Flory model, some of which are more selective [175,178].

Of closer relevance to the water gas shift are the nucleophilic

activation of CO to reduction of hydrogen [179], analogous to the
nucleophilic attack by water or hydroxide proposed for the water gas
shift. Coadsorption of CO and H2 on iron [180] and ruthenium [181]
leads to lessened dissociation for both. Most likely, the reaction
of water with surface carbon to give methane [182] and the interac-
tions among surface carbon, CO, and C02 [183] are relevant to both
the water gas shift and Fischer-Tropsch. However, more information
is needed on the elementary reactions.

Two of the components of the water gas equilibrium, CO and H2,

are the starting materials for the Fischer-Tropsch synthesis. The
other two, C02 and H20, are products of the synthesis. However, the
Fischer-Tropsch reactants are on opposite sides of the water gas
equilibrium, as are the products. Thus, the water gas shift can help
or hinder the Fischer-Tropsch synthesis by altering the concentra-
tions of reactants and products.

Since the two processes have these components in common, adsorp-

tion and desorption reactions must be shared, as must the dissocia-
tion reactions of H2, H20, and C02. Aside from these reactions,
there appear to be no further overlaps between the water gas shift
and Fischer-Tropsch. The dissociation of CO appears not to be a part
of the water gas shift. Hydroxycarbonyl may be an intermediate in
acid or ester formation in Fischer-Tropsch, but evidence here is very
limited. Also, if formate intermediates, which have been observed
under Fischer-Tropsch conditions, are hydrogenated significantly,
some of these reactions may overlap.

If the overlaps between the water gas shift and Fischer-Tropsch

are limited to adsorption-desorption and association-dissociation
reactions, then consideration of modifications to Fischer-Tropsch
catalysts to provide oxygen rejection by H20 rather than CO 2 may be
relatively simple. Catalysts on which the H20 association and de-
sorption reactions are favored over the C02 association and desorp-
tion reactions will give a water product. Unfortunately, data for
these reactions is not available for a wide range of metals.

Further applications of the relation between the water gas shift

and Fischer-Tropsch are in understanding the KOlbel-Englehardt syn-
thesis and C02 methanation. The differences here are not in the
elementary reactions, but in the relative concentrations of reactants

and intermediates, which may give rise to different paths by effec-

tively reducing the rate of one path to zero, while making signifi-
cant another path that is inactive in the water gas shift or Fischer-

The overlap, however, of the association-dissociation and ad-

sorption-desorption reactions between the water gas shift and Fisch-
er-Tropsch means that it is not useful to separate these two pro-
cesses conceptually. The water gas shift, although a simpler set of
elementary reactions than Fischer-Tropsch, cannot itself be regarded
as an elementary reaction. An attempt at a mathematical separation
of the kinetics of the water gas shift from the kinetics of Fischer-
Tropsch will give meaningless results because the sets of elementary
reactions are not orthogonal. Thus, the shorthand of attributing
various effects observed in the Fischer-Tropsch synthesis to the
water gas shift should be avoided, and an attempt should be made to
attribute the effects to a particular elementary reaction or set of
elementary reactions.


Although the scope of this paper does not permit a detailed dis-
cussion of methanol synthesis, that reaction must be closely related
to the water gas shift over the low-temperature catalyst, since the
catalysts are similar. An ideal situation would be to be able to
write a list of elementary reactions encompassing both methanol syn-
thesis and the water gas shift. However, the intermediates in meth-
anol synthesis have not been unambiguously identified [184]. Clearly
the largest difference between the two processes is in the oxidizing
power of the reactants: synthesis gas for methanol is more reducing
than either side of the water gas equilibrium. However, small
amounts of C02 promote methanol synthesis. Again, as in the Fischer-
Tropsch synthesis, adsorption-desorption of HZ, HZO, CO, and COZ and
association-dissociation reactions of HZ, H20, and C02 are probably
common to both processes. The co that is hydrogenated to methanol
remains associated, as is likely for the CO that is oxidized in the
water gas shift. It is not clear whether the intermediates in co hy-
drogenation are bonded to the catalyst through oxygen (formate-meth-
oxy route) or through carbon (formyl-hydroxymethyl route), or whether
both of these routes are significant [185].


The water gas shift shares a significant number of elementary

reactions with the Fischer-Tropsch synthesis, and probably with the
Kolbel-Engelhardt synthesis, C02 methanation, and methanol synthesis.
Oxygen transfer from C02 or H20 to the catalyst and then to H2 or CO
may be the predominant route at higher temperatures in heterogeneous
systems and appears to be possible for homogeneous systems. However,
a C-H-O intermediate also participates at lower temperatures, par-
ticularly in homogeneous systems. This intermediate may be formate
or hydroxycarbony1, but little information now exists that can clar-
ify whether one or both are important.

More information is needed about the interactions of H2, H20,

CO, and C02 on well-characterized oxide surfaces. Studies of formic
acid as a model compound may also be useful, but caution is necessary
in their interpretation. The relationship of homogeneous reactions
to heterogeneous catalysis is still not clear, but the parallels in
the water gas shift appear to be significant. Again, more studies
of the intermediates on surfaces will help to elucidate this rela-


I am grateful to R. G. Herman for suggesting the topic of this

paper and to G. L. DePoorter for helpful discussions. This work was
supported by the United States Department of Energy.


1. C. K. Rofer-DePoorter, Chem. Rev., 81:447 (1981).

2. Y. Kim, H. C. Peebles, and J. M. White, Surf. Sci., 114:363
3. G. Ert1, S. B. Lee, and M. Weiss. Surf. Sci., 111:L711 (1981).
4. E. I. Ko and R. J. Madix, Surf. Sci., 109:221 (1981)
5. M. Kiskinova and D. W. Goodman, Surf. Sci., 109:L555 (1981).
6. a) F. P. Netzer and T. E. Madey, J. Chem. Phys., 76:710 (1982).
b) K. E. Foley and N. Winograd, Surf. Sci., 116:1 (1982).
7. R. M. Kroeker and J. Pacansky, J. Chem. Phys., 76:3291 (1982).
8. M. D. Baker, N. D. S. Canning, and M. A. Chesters, Surf. Sci.,
111:452 (1981).
9. F. Solymosi and A. Erdohe1yi, Surf. Sci., 110:L630 (1981).
10. Z. Muryama, I. Kojima, E. Miyazaki, and I. Yasumori, Surf. Sci.,
118:L281 (1982).
11. M. R. McClellan, J. L. Gland, and F. R. McFee1ey, Surf. Sci.,
112:63 (1981).
12. E. L. Garfunke1, J. E. Crowell, and G. A. Somorjai, J. Phys.
Chem., 86:310 (1982).
13. M. P. Kiskinova, Surf. Sci., 111:584 (1981).
14. C. M. A. M. Mesters, A. F. H. Wie1ers, 0. L. J. Gijzeman, J. W.
Geus, and G. A. Bootsma, Surf. Sci., 115:237 (1982).
15. H. I. Lee, B. E. Koe1, W. M. Daniel, and J. M. White, J. Cata1.,
74:192 (1982).
16. a) T. Iizuka and Y. Tanaka, J. Cata1., 70:449 (1981).
b) F. Solymosi and A. Erdohe1yi, J. Cata1., 70:451 (1981) and
references therein.

17. F. Solymosi, A. ErdBhe1yi, and T. Bansagi, J. Cata1., 68:371

(1981) •
18. P. MUnster and H. J. Grabke, Arch. EisenhUttenwes., 51:319
(1980) •
19. a) M. Masuda, J. Res. Inst. Cata1., Hokkaido Univ., 14:85
b) M. Masuda and K. Miyahara, Bull. Chern. Soc. Jpn., 47:1058
c) M. Masuda, J. Res. Inst. Cata1., Hokkaido Univ., 24:83 (1977).
20. D. C. Grenoble, M. M. Estadt, and D. F. Ollis, J. Cata1., 67:90
21. a) M. Niwa, T. Iizuka, and J. H. Lunsford, J. Chern. Soc., Chern.
Commun., 684 (1979).
b) M. Niwa and J. H. Lunsford, J. Cata1., 75:302 (1982).
22. B. L. Gustafson and J. H. Lunsford, J. Cata1., 74:393 (1982).
23. a) J. Horiuti and T. Nakamura, Adv. Cata1., 17:1 (1967).
b) J. Horiuti, Ann. N. Y. Acad. Sci., 213:5 (1973).
24. R. P. Eischens, W. A. P1iskin, and M. J. D. Low, J. Cata1., 1:
180 (1962).
25. A. L. Dent and R. J. Kokes, J. Phys. Chern., 73:3772 (1969).
26. F. Boccuzzi, E. Borello, A. Zecchina, A. Bossi, and M. Camia,
J. Cata1., 51:150 (1978).
27. G. L. Griffin and J. T. Yates, Jr., J. Chern. Phys., 77:3744
(1982) •
28. a) C. C. Chang and R. J. Kokes, J. Am. Chern. Soc., 93:7107
(1971) •
b) C. C. Chang, L. T. Dixon, and R. J. Kokes, J. Phys. Chem.,
77:2634 (1973).
29. W. C. Conner, Jr., and R. J. Kokes, J. Cata1., 36:199 (1975).
30. A. Baranski and J. Ga1uzka, J. Cata1., 44:259 (1976).
31. A. Baranski and R. J. Cvetanovic, J. Phys. Chem., 75:208 (1971).
32. D. Narayana, V. S. Subrahmanyarn, J. La1, M. M. Ali, and V.
Kesavu1u, J. Phys. Chern., 74:779 (1970).
33. G. L. Griffin and J. T. Yates, Jr., J. Cata1., 73:396 (1982).
34. Yu. P. Borisevich, Yu. V. Fornichev, and M. E. Levinter, Zh. Fiz.
Khim., 55:2149 (1981); Russ. J. Phys. Chern., 55:1222 (1981).
35. R. L. Burwell, Jr., and K. S. Stec, J. ColI. Interface Sci.,
58: 54 (1977).
36. P. P. M. M. Wittgen, C. Groeneveld, J. H. G. J. Janssens, M. L.
J. A. Wetze1s, and G. C. A. Schuit, J. Cata1., 59:168 (1979).
37. T. Morimoto. Y. Yokota, and M. Nagao, J. Colloid Interface Sci.,
64:188 (1978).
38. a) E. McCafferty, V. Pravdic, and A. C. Zettlemoyer, Trans.
Faraday Soc., 66:1720 (1970).
b) E. McCafferty and A. C. Zettlemoyer, Discuss. Faraday Soc.,
52:239 (1971).
39. a) G. B1yho1der and E. A. Richardson, J. Phys. Chern., 66:2597
b) C. H. Rochester and S. A. Topham, J. Chern. Soc., Faraday
Trans. 1, 75:1073 (1978).

40. D. J. Dwyer, S. R. Kelemen, and A. Kaldor, J. Chern. Phys.,

76:1832 (1982).
41. N. A. Rubene, A. A. Davydov, A. V. Kravtsov, N. V. Usheva, and
S. I. Srnol'yaninov, Kinet. Katal., 17:465 (1976); Engl.
transl. 17:400 (1976).
42. a) M. Nagao, J. Phys. Chern., 75:3822 (1971).
b) T. Morimoto and M. Nagao, J. Phys. Chern., 78:1116 (1974).
43. G. Mattrnann, H. R. Oswald, and F. Schweizer, Helv. Chim. Acta,
55:1249 (1972).
44. a) C. T. Au, J. Breza, and M. W. Roberts, Chern. Phys. Lett.,
66: 340 (1979).
b) C. T. Au and M. W. Roberts, Chern. Phys. Lett., 74:472 (1980).
c) C. T. Au, M. W. Roberts, and A. R. Zhu, Surf. Sci., l15:Ll17
(1982) •
45. G. I. Salornatin, V. S. Sobolevskii, V. V. Grigor'ev, L. I.
Lafer, and V. I. Yakerson, Izv. Akad. Nauk SSSR, Sere Khirn.,
2437 (1979); Engl. transl., 2254 (1979).
46. a) E. Borello, G. Della Gatta, B. Fubini, C. Morterra, and G.
Venturello, J. Catal., 35:1 (1974).
b) G. Della Gatta, B. Fubini, and G. Venturello, Conf. IntI.
Thermodyn. Chirn., [C.R.], 4th, 7:72 (1975).
c) B. Fubini, G. Della Gatta, and G. Venturello, J. Colloid
Interface Sci., 64:470 (1978).
d) B. Fubini, E. Giarnello, and G. Della Gatta, J. Chirn. Phys.
Phys.-Chirn. BioI., 75:578 (1978).
47. a) R. P. Eischens and W. A. Pliskin, Adv. Catal., 10:1 (1958).
b) A. V. Kiselev and V. I. Lygin, Usp. Khirn., 31:351 (1962);
Russ. Chern. Rev., 31:175 (1962).
c) K. Klier, J. H. Shen, and A. C. Zettlemoyer, J. Phys. Chern.,
77:1458 (1973).
d) B. A. Morrow, I. A. Cody, and L. S. M. Lee, J. Phys. Chern. ,
80:2761 (1976) •
e) A. Nonaka and E. Ishizaki, J. Colloid Interface Sci., 62:381
(1977) •
f) A. J. van Roosrnalen and J. C. Mol, J. Phys. Chem. , 83:2485
(1979) •
g) P. A. Sermon, J. Chern. Soc., Faraday Trans. 1, 76:885 (1980).
48. A. Zecchina, S. Coluccia, E. Guglielminotti, and G. Ghiotti,
J. Phys. Chern., 75:2774 (1971).
49. a) T. Morimoto, M. Nagao, and J. Irnai, Bull. Chern. Soc. Jpn.,
44:1282 (1971).
b) T. Morimoto, M. Nagao, and J. Irnai, Bull. Chern. Soc. Jpn.,
47:2994 (1974).
50. a) P. G. Harrison and E. W. Thornton, J. Chern. Soc., Faraday
Trans. 1, 74:2703 (1978).
b) P. G. Harrison and E. W. Thornton, J. Chern. Soc., Faraday
Trans. 1, 75:1487 (1978).
51. B. Rebenstorf and R. Larsson, Acta. Chern. Scand. A, 34:239
52. E. A. Paukshtis, R. I. Soltanov, and E. N. Yurchenko, React.
Kinet. Cata1. Lett., 16:93 (1981).

53. K. H. Kim, H. S. Han, and J. S. Cho, J. Phys. Chern., 83:1286

54. G. M. Kozub, I. G. Voroshi1ov, L. M. Roev, and M. T. Rusov,
Kinet. Kata1., 17:1040 (1976).
55. S. R. Kelemen, A. Ka1dor, and D. J. Dwyer, Surf. Sci., 121:45
(1982) •
56. w. Hert1 and R. J. Farrauto, J. Cata1., 29:352 (1973).
57. G. I. Sa1omatin, L. I. Lafer, and V. I. Yakerson, Izv. Akad.
Nauk SSSR, Sere Khim., 1445 (1979); Engl. trans1., 1349
(1979) •
58. J. W. London and A. T. Bell, J. Cata1., 31:32 (1973).
59. A. A. Davydov, N. A. Rubene, and A. A. Budneva, Kinet. Kata1.,
19:969 (1978); Engl. trans1., 19:776 (1979).
60. R. A. Gardner and R. H. Petrucci, J. Am. Chem. Soc., 82:5051
61. D. A. Seanor and C. H. Amberg, J. Chern. Phys., 42:2967 (1965).
62. K. P. de Jong, J. W. Geus, and J. Joziasse, J. Cata1., 65:437
(1980) •
63. M. Pasqua1i, C. F1oriani, G. Venturi, A. Gaetani-Manfredotti,
and A. Chiesi-Vi11a, J. Am. Chem. Soc., 104:4092 (1982).
64. J. H. Taylor and C. H. Amberg, Can. J. Chem., 39:535 (1961).
65. C. H. Amberg and D. A. Seanor, Proc. 3rd Intern. Congr. Cata1.,
450 (1965).
66. I. G. Voroshi1ov, L. M. Roev, G. M. Kozub, N. K. Lunev, N. A.
Pav1ovskii, and M. T. Rusov, Teor. Eksp. Khim., 11:256
(1975); Engl. trans1., 11:212 (1975).
67. W. Hotan, W. GBpe1, and R. Haul, Surf. Sci., 83:162 (1979).
68. a) M. J. Sayers, M. R. McClellan, R. R. Gay, E. I. Solomon, and
F. R. McFeely, Chern. Phys. Lett., 75:575 (1980).
b) R. R. Gay, M. H. Nodine, V. E. Henrich, H. J. Zeiger, and
E. I. Solomon, J. Am. Chem. Soc., 102:6752 (1980).
c) M. R. McClellan, M. Trenary, N. D. Shinn, M. J. Sayers, K. D.
D'Amico, E. I. Solomon, and F. R. McFeely, J. Chem. Phys.,
74:4726 (1981).
69. L. H. Little and Ch. H. Amberg, Can. J. Chem., 40:1997 (1962).
70. D. D. E1ey, C. H. Rochester, and M. S. Scurre11, J. Cata1.,
29: 20 (1973).
71. E. Borello, A. Zecchina, C. Morterra, and G. Ghiotti, J. Phys.
Chern., 73:1286 (1969).
72. J. Rask6 and F. Solymosi, Acta Chim., Acad. Sci. Hung., 95:389
(1977) •
73. A. Clark, J. N. Finch, and B. H. Ashe, Proc. Int. Congr. Cata1.,
3rd 1964:1010 (Pub. 1965).
74. M. Kobayashi, T. Date, and H. Kobayashi, Bull. Chem. Soc. Jpn.,
49:3014 (1976).
75. N. D. Parkyns, J. Chem. Soc. (A), 1910 (1967).
76. G. Della Gatta, B. Fubini, G. Ghiotti, and C. Morterra,
J. Cata1., 43:90 (1976).
77. A. Stanislaus, M. J. B. Evans, and R. F. Mann, J. Phys. Chern.,
76:2349 (1972).

78. B. W. Krupay and Y. Amenomiya, J. Cata1., 67:362 (1981).

79. a) H. Ko1be1, M. Ra1ek, and P. Jiru, Z. Naturforsch. A, 25:670
(1970) •
b) R. St. C. Smart, T. L. Slager, L. H. Little, and R. G.
Greenier, J. Phys. Chem., 77:1019 (1973).
c) R. J. Breakspere, M. A. A1-Dugaither, Aust. J. Chem., 30:241
80. a) J. H. Lunsford and J. P. Jayne, J. Chem. Phys., 44:1492
b) v. E. Shubin, V. A. Shvets, K. Dyrek, K. Moca1a, and V. B.
Kazansky, React. Kinet. Cata1. Lett., 14:239 (1980).
81- D. Cordischi, V. Indovina, and M. Occhiuzzi, J. Chem. Soc.,
Faraday Trans. I, 76:1147 (1980).
82. S. Co1uccia, E. Garrone, E. Gug1ie1minotti, and A. Zecchina,
J. Chem. Soc., Faraday Trans. I, 77:1063 (1981).
83. K. Atherton, G. Newbold, and J. A. Hockey, Discuss. Faraday
Soc., 52:33 (1971).
84. G. Busca and V. Lorenze11i, Mat. Chem., 5:213 (1980).
85. a) A. Zecchina, C. Morterra, G. Ghiotti, and E. Borello, J.
Phys. Chem., 73:1292 (1969).
b) A. Zecchina, S. Co1uccia, E. Gug1ie1minotti, and G. Ghiotti,
J. Phys. Chem., 75:2790 (1971).
86. V. G. Amerikov and L. A. Kasatkina, Kinet. Kata1., 12:165
(1971); Engl. trans1., 12:137 (1971).
87. S. Matsushita and T. Nakata, J. Chem. Phys., 36:665 (1962).
88. A. Ueno and C. O. Bennett, J. Cata1., 54:31 (1978).
89. G. Hochstrasser and J. F. Antonini, Surf. Sci., 32:644 (1972).
90. J. F. Antonini and G. Hochstrasser, Surf. Sci., 32:665 (1972).
91- a) S. J. Gregg and J. D.F. Ramsay, J. Phys. Chem., 73:1243
b) E. Baumgarten and A. Zachos, Spectrochim. Acta, 37A:93
92. a) N. D. Parkyns, J. Chem. Soc. (A), 410 (1969).
b) C. Morterra, S. Co1uccia, G. Ghiotti and A. Zecchina, Z.
Phys. Chem. N. F., 104:275 (1977).
c) J. B. Peri, J. Phys. Chem., 70:3168 (1966).
93. C. Morterra, A. Zecchina, S. Co1uccia, and A. Chiorino, J. Chem.
Soc., Faraday Trans. I, 73:1544 (1977).
94. N. D. Parkyns, J. Phys. Chem., 75:526 (1971).
95. S. T. King, App1. Spectrosc., 34:632 (1980).
96. S. J. Gregg and J. D. Ramsay, J. Chem. Soc. (A), 2784 (1970).
97. M. P. Rosynek, J. Phys. Chem., 79:1280 (1975).
98. S. Gambarotta, F. Arena, C. F1oriani, and P. F. Zanazzi, J. Am.
Chem. Soc., 104:5082 (1982).
99. J. V. Evans and T. L. Whate1ey, Trans. Faraday Soc., 63:2769
(1967) •
100. J. Suh, M. Cheong, and M. P. Suh, J. Am. Chem. Soc., 104:1654
101. a) A. Ueno, T. Yamamoto, T. Onishi, and K. Tamaru, Bull. Chem.
Soc. Jpn., 42:3040 (1969).

b) A. Ueno, T. Onishi, and K. Tamaru, Trans. Faraday Soc.,

66:756 (1970).
c) K. Tamaru and T. Onishi, App1. Spectrosc. Rev., 9:133
102. E. Ramaroson, R. Kieffer, and A. Kiennemann, App1. Cata1.,
4:281 (1982).
103. A. De1uzarche, R. Kieffer, and M. Papadopoulos, C. R. Hebd.
Seances Acad. Sci., Sere C, 287:25 (1978).
104. Ya. M. Grigor'ev, D. v. Pozdnyakov, and V. N. Fi1imonov, Zh.
Fiz. Khim., 46:316 (1972); Russ. J. Phys. Chern., 46:18-6--
(1972) •
105. F. S. Stone, Adv. Cata1., 13:1 (1962).
106. A. Zecchina, G. Ghiotti, C. Morterra, and E. Borello, J. Phys.
Chern., 73:1295 (1969).
107. A. A. Davydov, Yu. M. Shchekochikhin, and N. P. Keier, Kinet.
Katal., 10:l341 (1969); Engl. transl., 10:1103 (1969).
108. T. S. Nargarjunan, M. V. C. Sastri, and J. C. Kuriacose,
J. Cata1., 2:223 (1963).
109. a) S. Tsuchiya and T. Shiba, J. Cata1., 4:116 (1965).
b) S. Tsuchiya and T. Shiba, Bull. Chern. Soc. Jpn., 38:1726
c) S. Tsuchiya and T. Shiba, Bull. Chern. Soc. JEn. , 40:1086
d) S. Tsuchiya and T. Shiba, Bull. Chern. Soc. JEn. , 41:573
110. C. Aharoni and F. C. Tompkins, Trans. Faradal Soc. , 66:434
(1970) •
111. F. Boccuzzi, E. Garrone, A. Zecchina, A. Bossi, and M. Camia,
J. Cata1., 51:160 (1978).
112. E. Giame110 and B. Fubini, React. Kinet. Cata1. Lett., 16:355
113. G. L. Griffin and J. T. Yates, Jr., J. Chern. Phls., 77:3751
114. D. S. Newsome, Cata1. Rev.-Sci. Eng., 21:275 (1980).
115. G. G. Shchibrya, N. M. Morozov, and M. I. Temkin, Kinet.
Kata1., 6:1057 (1965); Engl. trans1., 6:955 (1965).
116. S. Stotz, Ber. Bunsenges., 70:37 (1966).
117. P. J. Meschter and H. J. Grabke, Met. Trans. B, 10B:323 (1979).
118. a) S. Oki and R. Mezaki, J. Phls. Chern., 77:447 (1973).
b) S. Oki and R. Mezaki, J. Phls. Chern., 77:1601 (1973).
c) R. Mezaki and S. Oki, J. Cata1., 30:488 (1973).
d) W. F. Podolski and Y. G. Kim, Ind. Eng. Chern., Process Des.
Dev., 13:415 (1974).
e) P. Fott, J. Voso1sobe and V. Glaser, ColI. Czech. Chern.
Commun., 44:652 (1979).
119. a) T. M. Yur'eva, G. K. Boreskov, and V. She Gruver, Kinet.
Kata1., 10:862 (1969); Engl. trans1., 10:705 (1969).
b) C. Aharoni and H. Starer, Can. J. Chern., 52:4044 (1974).
c) V. V. Grivor'ev, V. N. Gel 'man, V. S. Sobo1evskii, A. I.
Kreinde1', E. Z. Go1osman, G. I. Sa1omatin, G. A. Dantsig,

T. R. Abdu11aev, L. I. Lafer, and V. I. Yakerson, Izv. Akad.

Nauk SSSR, Sere Khim., 1168 (1977); Engl. trans1., 1015
(1977) •
d) A. A. Davydov, G. K. Boreskov, T. M. Yur'eva, and N. A.
Rubene, Dok1. Akad. Nauk SSSR, 236:1402 (1977); Dok1ady
Phys. Chern., 236:1018 (1977).
120. a) T. van Herwijnen and W. A. de Jong, J. Cata1., 63:83 (1980).
b) T. van Herwijnen, R. T. Gucza1ski, and W. A. de Jong, J.
Cata1., 63:94 (1980). --
121. P. N. Mukherjee, P. K. Basu, and S. K. Roy, J. Indian Chern.
Soc., 55:1204 (1979).
122. a) Y. Amenorniya, Appl. Spectrosc., 32:484 (1978).
b) Y. Amenorniya, J. Cata1., 55:205 (1978).
c) Y. Amenorniya, J. Catal., 57:64 (1978).
d) Y. Amenorniya and G. P1eizier, J. Cata1., 76:345 (1982).
123. a) C. H. Cheng, D. E. Hendricksen, and R. Eisenberg, J. Am.
Chern. Soc., 99:2791 (1977).
b) E. C. Baker, D. E. Hendricksen, and R. Eisenberg, J. Am.
Chern. Soc., 102:1020 (1980).
124. P. C. Ford, P. Yarrow, and H. Cohen, ACS Symp. Ser., 152:95
125. V. N. Zudin, V. A. Likho1obov, Yu. I. Ermakov, and N. K. Ere-
rnenko, Kinet. Kata1., 18:524 (1977); Engl. trans1., 18:440
126. a) R. B. King, C. C. Frazier, R. M. Hanes, and A. D. King, Jr.,
J. Am. Chern. Soc., 100:2925 (1978).
b) C. C. Frazier, R. M. Hanes, A. D. King, Jr., and R. B. King,
Adv. Chern. Ser., 173:94 (1979).
c) R. B. King, A. D. King, Jr., and D. B. Yang, ACS Symp. Ser.,
152:123 (1981).
d) A. D. King, Jr., R. B. King, and D. B. Yang, J. Am. Chern.
Soc., 103:2699 (1981).
127. A. D. King, Jr., R. B. King, D. B. Yang, J. Am. Chern. Soc.,
102:1028 (1980).
128. a) R. M. Laine, R. G. Rinker, and P. C. Ford, J. Am. Chern.
Soc., 99:252 (1977).
b) ~. Ford, C. Ungermann, V. Landis, S. A. Moya, R. C.
Rinker, and R. M. Laine, Adv. Chern. Ser., 173:81 (1979).
c) C. Ungermann, V. Landis, S. A. Moya, H. Cohen, H. Walker,
R. G. Pearson, R. G. Rinker, and P. C. Ford, J. Am. Chern.
Soc., 101:5922 (1979).
129. P. Giannoccaro, G. Vasapo110, and A. Sacco, J. Chern. Soc.,
Chern. Commun., 1136 (1980).
130. R. G. Pearson and H. Mauermann, J. Am. Chern. Soc., 104:500
13l. P. C. Ford, Acc. Chern. Res., 14:31 (1981).
132. W. A. R. Slegir, R. S. Sapienza, and B. Easterling, ACS Symp.
Ser., 152:325 (1981). .
133. J. C. Bricker, C. C. Nagel, and S. G. Shore, J. Am. Chern. Soc.,
104:1444 (1982).

134. R. G. Nuzzo, D. Feit1er, and G. M. Whitesides, J. Am. Chern.

Soc., 101:3683 (1979).
135. K. Kaneda, M. Hiraki, K. Sano, T. Imanaka, and S~ Teranishi,
J. Mol. Cata1., 9:227 (1980).
136. a) Y. Yoshida, T. Okano, and S. Otsuka, ACS Symp. Ser., 152:95
(1981) •
b) T. Yoshida, T. Okano, and S. Otsuka, J. Am. Chem. Soc.,
103:3411 (1981).
137. a) H. C. Kang, C. H. Mauldin, T. Cole, W. Slegir, K. Cann, and
R. Pettit, J. Am. Chern. Soc., 99:8323 (1977).
b) R. Pettit, K. Cann, T. Cole, and C. H. Mauldin, Adv. Chern.
Ser., 173:121 (1979).
138. T. Yoshida, Y. Ueda, and S. Otsuka, J. Am. Chern. Soc., 100:3941
139. c. H. Cheng and R. Eisenberg, J. Am. Chern. Soc., 100:5968
140. a) B. R. James, G. L. Rempel, and F. T. T. Ng, J. Chern. Soc.
j!L, 2454 (1969).
b) D. J. Darensbourg, M. Y. Darensbourg, N. Walker, J. A. Froe-
lich, and H. L. C. Barros, Inorg. Chern., 18:1401 (1979).
c) D. J. Darensbourg, M. Y. Darensbourg, R. R. Burch, Jr., J.
A. Froelich, and M. J. Incorvia, Adv. Chern. Ser., 173:106
(1979) •
d) D. J. Darensbourg, B. J. Baldwin, and J. A. Froelich, J. Am.
Chem. Soc., 102:4688 (1980).
e) D. J. Darensbourg and A. Rokicki, ACS Symp. Ser., 152:107
141. N. Crice, S. C. Kao, and R. Pettit, J. Am. Chern. Soc., 101:1627
(1979) •
142. a) D. J. Darensbourg, A. Rokicki, and M. Y. Darensbourg, J. Am.
Chern. Soc., 103:3223 (1981).
b) D. J. Darensbourg and A. Rokicki, J. Am. Chern. Soc., 104:349
143. s. Atta1i, R. Mathieu, and G. J. Leigh, J. Mol. Cata1., 14:293
(1982) •
144. A. D. King, Jr., R. B. King, and E. L. Sailers, III, J. Am.
Chern. Soc., 103:1867 (1981).
145. D. J. Darensbourg, M. B. Fischer, R. E. Schmidt, Jr., and B. J.
Baldwin, J. Am. Chern. Soc., 103:1297 (1981).
146. T. Yoshida, T. Okano, and S. Otsuka, J. Am. Chern. Soc., 102:
5966 (1980).
147. V. N. Zudin, V. A. Likho1obov, V. M. Mastikhin, O. B. Lap ina ,
and Yu. I. Ermakov, Koord. Khim., 5:432 (1979); Engl.
trans1., 5:330 (1979).
148. D. R. Gard and T. L. Brown, J. Am. Chern. Soc., 104:6340 (1982).
149. D. J. Cole-Hamilton, J. Chern. Soc., 1213 (1980).
150. J. M. Maher and N. J. Cooper, J. Am. Chern. Soc., 102:7604
(1980) •
151. a) J. J. Verdonck, P. A. Jacobs, and J. B. Uytterhoeven, J.
Chem. Soc., Chern. Commun., 181 (1979).
b) J. J. Verdonck, R. A. Schoonheydt, and P. A. Jacobs, Proc.
Int. Congr. Catal., 7th, 1980, 911 (Pub. 1981).
152. J. M. Maher, G. R. Lee, and N. J. Cooper, J. Am. Chern. Soc.,
104:6797 (1982).
153. a) N. M. D. Brown, R. B. Floyd, and D. G. Walmsley, J. Chern.
Soc., Faraday Trans. 1, 75:17 (1979).
b) R. Magno and J. G. Adler, J. App1. Phys., 49:4465 (1978).
c) D. G. Walmsley, W. J. Nelson, N. M. D. Brown, and R. B.
Floyd, App1. Surf. Sci., 5:107 (1980).
d) J. A. Reimer and R. W. Vaughan, J. Mag. Res., 41:483 (1980).
e) T. M. Duncan and R. W. Vaughan, J. Cata1., 67:49 (1981).
f) T. M. Duncan and R. W. Vaughan, J. Cata1., 67:469 (1981).
g) G. I. Sa10matin, V. S. Sobo1evskii, L. I. Lafer, and V. I.
Yakerson, Izv. Akad. Nauk SSSR, Sere Khim., 1988 (1980);
Engl. trans1., 1405 (1980).
h) G. Busca and V. Lorenze11i, J. Cata1., 66:155 (1980).
i) A. K. Bhattacharya, J. Chem. Soc., Faraday Trans. 1, 75:863
j) I. I. Bobrova, E. A. Paukshtis, E. N. Yurchenko, V. A. Saz-
onov, and V. V. Popovskii, Kinet. Kata1., 22:696 (1981).
k) R. W. Joyner and M. W. Roberts, Proc. Roy. Soc. London A,
350:107 (1976).
1) B. A. Sexton, Surf. Sci., 88:319 (1979).
m) S. A. Isa, R. W. Joyner, M. H. Mat100b, and M. W. Roberts,
App1. Surf. Sci., 5:345 (1980).
n) M. Mohri, M. Hashiba, K. Watanabe and T. Yamashina, z. Phys.
Chem. N. F. 109:233 (1978).
0) M. Mohri, H. Kakibayashi, K. Watanabe, and T. Yamashina,
App1. Surf. Sci., 1:170 (1978).
154. a) D. E. Tevau1t and M. C. Lin, NBS Spec. Pub1. (U.S.) 561-2,
1551-1568 (1979).
b) D. E. Tevau1t, M. C. Lin, M. E. Umstead, and R. R. Smard-
zewski, Int. J. Chem. Kinet., 11:445 (1979).
155. a) S. W. Johnson and R. J. Madix, Surf. Sci., 66:189 (1977).
b) D. H. S. Ying and R. J. Madix, Inorg. Chem., 17:1103 (1978).
c) L. A. Larson and J. T. Dickinson, Surf. Sci., 84:17 (1979).
156. M. Ito and W. Suetaka, J. Cata1., 54:13 (1978).
157. S. H. Strauss, K. H. Whitmire, and D. F. Shriver, J. Organomet.
Chem., 174:C59 (1979).
158. H. P. Bonze1 and H. J. Krebs, Surf. Sci., 117:639 (1982).
159. a) K. Kuchynka, J. Fusek, and O. Strouf, ColI. Czech. Chem.
Commun., 46:2328 (1981).
b) O. Strouf, K. Kuchynka, and J. Fusek, ColI. Czech. Chem.
Commun. 46:2336 (1981).
160. J. T. Yates, Jr., and R. R. Cavanagh, J. Cata1., 74:97 (1982).
161. a) J. A. G1adsyz, Adv. Organomet. Chem., 20:1 (1982).
b) C. P. Casey, M. A. Andrews, D. R. McAlister, W. D. Jones,
and S. G. Harsy, J. Mol. Cata1., 13:43 (1981).
c) R. C. Schoening, J. L. Vidal, and R. A. Fiato, J. Mol.
Cata1., 13:83 (1981).

d) J. R. Sweet and W. A. G. Graham, J. Am. Chem. Soc., 104:2811

e) B. B. Wayland, B. A. Woods, and R. Pierce, J. Am. Chem.
Soc., 104:302 (1982).
162. a) K. H. Whitmire and D. F. Shriver, J. Am. Chem. Soc., 103:
6754 (1981).
b) J. W. Ko1is, F. Baso10, and D. F. Shriver, J. Am. Chem.
Soc., 104:5626 (1982).
c) M. A. Drezdzon, K. H. Whitmire, A. A. Bhattacharyya, W. L.
Hsu, C. C. Nagel, S. G. Shore, and D. F. Shriver, J. Am.
Chem. Soc., 104:5630 (1982).
163. a) J. C. Hayes, G. D. N. Pearson, and N. J. Cooper, J. Am.
Chem. Soc., 103:4648 (1981).
b) D. L. Thorn and T. H. Tulip, J. Am. Chem. Soc., 103:5984
(1981) •
c) D. H. Berry, J. E. Bercaw, A. J. Jircitano, and K. B.
Mertes, J. Am. Chem. Soc., 104:4712 (1982). .
d) C. P. Casey and P. J. Fagan, J. Am. Chem. Soc., 104:4950
164. a) A. H. Janowicz and R. G. Bergman, J. Am. Chem. Soc., 104:352
b) J. K. Hoyano and W. A. G. Graham, J. Am. Chem. Soc., 104:
3723 (1982).
165. a) A. E. Stevens and J. L. Beauchamp, J. Am. Chem. Soc., 103:
190 (1981).
b) L. F. Halle, P. B. Armentrout, and J. L. Beauchamp, J. Am.
Chem. Soc., 103:962 (1981).
c) P. B. Armentrout, L. F. Halle, and J. L. Beauchamp, J. Am.
Chem. Soc., 103:6501 (1981).
d) G. D. Byrd, R. C. Burnier, and B. S. Freiser, J. Am. Chem.
Soc., 104:3565 (1982).
166. R. F. Jordan and J. R. Norton, J. Am. Chem. Soc., 104:1255
167. a) K. G. Cau1ton, J. Mol. Cata1., 13:71 (1981).
b) J. A. Marsella, K. Fo1ting, J. C. Huffman, and K. G. Cau1-
ton, J. Am. Chem. Soc., 103:5596 (1981).
c) R. S. Thre1ke1 and J. E. Bercaw, J. Am. Chem. Soc., 103:2650
(1981) •
d) J. H. Teuben, E. J. M. de Boer, A. H. Klazinga, and E. K1ei,
J. Mol. Cata1., 13:107 (1981).
168. a) E. A. Maata and T. J. Marks, J. Am. Chem. Soc., 103:3576
b) P. J. Fagan, J. M. Manriquez, E. A. Maata, A. M. Seyam, and
T. J. Marks, J. Am. Chem. Soc., 103:6650 (1981).
c) P. J. Fagan, K. G. Mo1oy, and T. J. Marks, J. Am. Chem.
Soc., 103:6959 (1981).
169. W. A. Herrmann, Angew. Chem. Int. Ed. Engl., 21:117 (1982).
170. a) A. O. I. Rautavuoma and H. S. van der Baan, App1. Cata1.,
1:247 (1981).
b) J. Happel, V. Fthenakis, I. Suzuki, T. Yoshida, and S.
Ozawa, Proc. Int. Congr. Cata1., 7th, 1980, 542 (Pub. 1981).

c) J. Happel, H. Y. Chen, M. Otarod, S. Ozawa, A. J. Severdia,

T. Yoshida, and V. Fthenakis, J. Cata1., 75:314 (1982).
d) J. Ga1uszka, J. R. Chang, and Y. Amenomiya, Proc. Int.
Congr. Cata1., 7th, 1980, 529 (Pub. 1981).
e) F. Solymosi, I. Tombacz, and M. Kocsis, J. Cata1., 75:78
(1982) •
f) N. W. Cant and A. T. Bell, J. Cata1., 73:257 (1982).
171. a) J. A. Da1mon and G. A. Martin, Proc. Int. Congr. Cata1.,
7th, 1980, 402 (Pub. 1981).
b) W. T. Osterloh, M. E. Cornell, and R. Pettit, J. Am. Chem.
Soc., 104:3759 (1982).
172. a) H. P. Bonze1 and H. J. Krebs, Surf. Sci., 109:L527 (1981).
b) J. W. Niemantsverdriet and A. M. van der Kraan, J. Cata1.,
72:385 (1981).
c) J. W. Niemantsverdriet, C. F. J. F1ipse, A. M. van der
Kraan, and J. J. van Loef, App1. Surf. Sci., 10:302 (1982).
d) J. P. Reymond, P. Meriaudeau, and S. J. Teichner, J. Cata1.,
75:39 (1982).
173. R. S. Polizzotti and J. A. Schwarz, J. CataI., 77:1 (1982).
174. P. R. Watson and G. A. Somorjai, J. Cata1., 74:282 (1982)
175. S. Novak, R. J. Madon, and H. Suh1, J. Cata1., 77:141 (1982).
176. R. B. Pannell, C. L. Kibby, and T. P. Kobylinski, Proc. Int.
Congr. Cata1., 7th, 1980, 447 (Pub. 1981).
177. R. J. Madon and W. F. Taylor, J.Cata1., 69:32 (1981).
178. S. Novak, R. J. Madon, and H. Suh1, J. Chem. Phys., 74:6083
(1981) •
179. K. M. Doxsee and R. H. Grubbs, J. Am. Chem. Soc., 103:7696
(1981) •
180. J. B. Benziger and R. J. Madix, Surf. Sci., 115:279 (1982).
18I. D. E. Peebles, J. A. Schriefe1s, and J. M. White, Surf. Sci.,
116:117 (1982).
182. S. E. Moore and J. H. Lunsford, J. Cata1., 77:297 (1982).
183. B. L. Gustafson and J. H. Lunsford, J. Cata1., 74:405 (1982).
184. a) H. H. Kung, Cata1. Rev.-Sci. Eng., 22:235 (1980).
b) K. Klier, Adv. Cata1., 31:243 (1982).
185. K. Klier, V. Chatikavanij, R. G. Herman, and G. W. Simmons,
J. Cata1., 74:343 (1982).

J. V. Bauer, B. W. Brian, S. A. Butter, P. N. Dyer, *

R. L. Parsons, and R. Pierantozzi

Air Products and Chemicals, Inc.

P. O. Box 538
Allentown, PA 18105


Energy R&D is undergoing considerable analysis in the U. S.

today, and the near term outlook for coal liquefaction is unclear.
However, in the longer term, coal is expected to become an economical
source of energy as petroleum and gas reserves dwindle further.
While oil accounts for about 10% of the world's recoverable fossil
fuel reserves, over two-thirds of these reserves are coal.

The production of liquid fuels from coal can be accomplished by

two distinct routes, direct and indirect liquefaction. Each route
offers the potential for high quality heating and transportation
fuels. The direct liquefaction processes add hydrogen to coal (usu-
ally donor solvent-mediated), while the indirect routes are two-step
processes involving the production of synthesis gas by coal gasifi-
cation, which is converted to liquid products in the second step.

The efficient conversion of the mixtures of carbon monoxide and

hydrogen (synthesis gas) produced by new generation coal gasifiers to
high quality fuels, such as diesel, is therefore of strategic impor-
tance. In 1980, a research program supported by the Department of
Energy was begun, to explore the development and evaluation of novel
slurry catalysts and reactor systems capable of selectively convert-
ing CO + H2 in a single-stage, liquid phase process [la]. A study of
the hydrodynamics of three-phase slurry reactors using cold flow mod-
elling techniques is an integral part of this program [lb].

The technical focus of this research coincides with the goals of

the DOE indirect liquefaction program by emphasizing the following

130 J. V. BAUER ET AL.


(a) The need for increased product selectivity. Novel cata-

lysts, where the Schulz-Flory product distribution limitation does
not hold, increase desirable liquid yields.

(b) Further integration of synthesis gas liquefaction with the

gasification step, by utilizing directly the CO-rich (CO/HZ > 1) syn-
thesis gas that is typical of newer generation gasifiers.

(c) The use of thermally more efficient synthesis gas conversion

reactors. Slurry phase reactors have advantages of improved temper-
ature control and heat recovery, high conversions and the promise of
using the new catalysts for processing high CO/HZ gases with improved
product quality.


A formal description of the chemistry of CO/HZ under Fischer-

Tropsch synthesis conditions involves the following generalized re-

ZnHZ + nCO ~ (-CHZ-)n + nHZO (~HR = -165 kJ) (1)

nHz + ZnCO ~ (-CHZ-)n + nCOz (~HR -Z05 kJ) (Z)

The difference between these two equations may be viewed as the
water-gas shift reaction (WGS),

HZO + CO t HZ + COZ (~HR = -39.7 kJ) (3)

A catalyst that allows the WGS reaction to proceed simultaneously

with the hydrocarbon growth reaction can, in effect, utilize the
higher ratio CO/HZ gas of equation (Z) by including the hydrogen
forming reaction (3) as a way of adjusting the CO/HZ ratio of the
synthesis gas.

For n-paraffin production from synthesis gas, the reaction stoi-

chiometry is shown by equations (4) and (5):

(Zn + l)HZ + nCO ~ CnHZn+Z + nHzO (4)

(n + I)HZ + ZnCO ~ CnHZn+Z + nCOz. (5)

In principle then, for reactions (1) and (Z), where olefins are prod-
ucts, CO/HZ utilization ratios may vary from 0.5 to Z. Where paraf-
fins are produced (e.g. (4) and (5», the CO/H2 usage ratio may vary
from n/(Zn + 1) to Zn/(n + 1), depending on whether the by-product
rejected is water or carbon dioxide. For example, for a paraffin

product with an average molecular weight of 15, the CO/H2 usage ratio
can vary from 0.48 to 1.88.

Literature on the Fischer-Tropsch synthesis is voluminous and

the most. previous research has involved fixed-bed, gas-solid reac-
tors. The early work was reviewed by Storch, Golumbic and Anderson
[2]. The more recent literature on Fischer-Tropsch synthesis in the
slurry phase has been reviewed by KBlbel and Ralek [3] and Poutsma
[4], where the potential incentives for using high CO/H2 gases (~1.5)
in liquid phase slurry reactors were pointed out. KBlbel also dis-
cussed the suitability of slurry systems to minimize carbon deposi-
tion on the catalyst, another advantage over fixed-bed processes,
particularly at high CO/H2 ratios. Other advantages include high
yields with minimum mass transfer limitations, high catalyst activi-
ty maintenance, and flexibility in product selectivity control.
Deckwer et al. have analyzed and modeled Fischer-Tropsch slurry re-
actor performance [5], while Satterfield et al. have examined the
literature on product distribution in Fischer-Tropsch slurry reac-
tors, particularly over iron catalysts [6]. It was. concluded that
product hydrocarbon selectivities follow a predicted Schulz-Flory
distribution characterized by the chain growth probability factor a,
and that previously reported deviations are probably due to experi-
mental artifacts. Fixed-bed reactor temperature gradients, insuffi-
cient time to reach steady-state, volatilization of certain frac-
tions, and condensation of high molecular weight material are some of
the experimental problems that can give misleading selectivity re-
sults. We have sought to avoid these previously encountered problems
by close attention to experimental analysis, and by determining prod-
uct distributions over long term, slurry phase operation with good
quantitative mass balances.


Stirred Slurry Reactors

The continuous, stirred, slurry phase reactor used for the

Fischer-Tropsch synthesis is based either on a 300 ml or a 1 liter
Autoclave Engineers unit, and is illustrated diagramatically in
Figure 1. Inlet CO and H2 streams are passed through separate oxygen
removal and drying stages, and for theCa stream, an additional iron
carbonyl removal stage using a heated alumina trap, before being
mixed and preheated. Using mass flow controllers, any desired ratio
of CO/H2 can be input to the stirred reactor system. The reactor is
fully baffled, and the gas inlet point is directly beneath the six,
flat-bladed impeller to maximize gas shear. The reactor operates in
a temperature range of 220-330°C, pressures of 1.1 to 3.5 MPa, and
space velocities up to a GHSV of 1000 hr-l. Products, together with
unreacted synthesis gas, are taken overhead through a heated partial
reflux condenser, maintained at a top temperature of about 200°C to
132 J. V. BAUER ET AL.




Figure 1. Continuous, automated Fischer-Tropsch slurry reactor

return vaporized slurry oil to the reactor.

The level of slurry in the reactor is continuously monitored by

determining the differential pressure (dp) between the gas inlet and
outlet streams, so that any build-up of higher molecular weight prod-
ucts that are not removed from the reactor with the exit gas flow can
be detected. When this occurs, hydrocarbon product can be removed
from the reactor directly via a heated sidestream, filtered through
a S ~m stainless steel sinter filter, and subjected to analysis.
Automatic slurry level control is possible via feedback from the dp
gauge, and a pump enables fresh or regenerated slurry catalyst to be
recycled back into the reactor. By determining the amount of slurry
oil withdrawn to maintain a constant level at a particular set of
process conditions, the higher molecular weight hydrocarbons that do
not distill with the gas phase product can be quantitatively included
in the material balance of the system. This procedure is essential
to obtain an overall product selectivity.

The product stream from the partial reflux condenser is led via
a heated line to a pressure reduction stage, and thence to a CS/C6
splitting column, to produce a condensed liquid phase and a gaseous
phase. The gaseous stream consists of unreacted CO/H2, C02 and prod-
ucts with carbon numbers from C1-CS. The condensed liquid product
consists of hydrocarbons with carbon numbers C6 and above and an
aqueous phase containing dissolved oxygenates. This method of prod-
uct collection avoids the use of high pressure traps, is more suited
to continuous operation, and, by reducing the number of product frac-

tions, is more accurate in obtaining material balances.

The whole system is designed to run continuously and automati-

cally when unattended, with automatic sampling of the split gas phase
stream and collection of liquid samples. The results presented in
this paper are all obtained during 21 day or longer continuous test
periods, to avoid transient or initial phenomena. After any change
in process parameters, the reactor system is allowed to equilibrate
for 14 to 16 hours before obtaining carbon and hydrogen material bal-
ances over an additional 8 hour period.

Because of the complexity of the Fischer-Tropsch product, equi-

libration of the reactor and the product collection systems, and a
flexible quantitative analysis scheme incorporating all products in-
cluding waxes, are required to produce good carbon and hydrogen ma-
terial balances and prevent misleading results. This procedure rou-
tinely produces material balances for C and H of 99 ± 1 to 3%.

Analytical and Data Handling System

Details of the analytical and computerized data handling system

have been published [7]. The scheme is illustrated diagramatically
in Figure 2, and consists of five separate gas chromatographs linked,
via a Sigma 10 computing integrator, to a Tektronix 40S2 microcom-
puter equipped with a 1.8 Mbyte disk system.

Unreacted synthesis gas, C02, and Cl to Cs isomers are analyzed

by a Carle 397B process gas chromatograph, while low concentrations
of C6+ in the gas phase product, due to inefficiencies in the CS/C6
splitting column, are analyzed by a Carle 111 GC with a Porapak QS
column. After separation and weighing of the condensed liquid
phases, aqueous phase samples are analyzed for Cl-C6 alcohols, alde-
hydes, ketones and acids using a 3 mm x 3 m 10% SP1200/l% H3P04 on
Chromo sorb WAW column. Samples of the separated organic phase are
analyzed for CS-C40 hydrocarbons using a 3 mm x 3 m SP2l00/Supelco-
port column, or, for additional resolution, an OV-lOOl WC 46 m x 0.2S
mm capillary column. Filtered wax samples removed directly from the
reactor are also analyzed on the 3 mm SP2l00 column for CS-C40 hydro-
carbons, and any contribution from the initial slurry oil is sub-
tracted from the analysis.

The analytical data is collected and temporarily stored in the

Sigma 10 integrator, before direct transfer to the Tektronix disc
system. After compilation into matrix format, the six data files for
each sample point are assembled by the computer into an overall prod-
uct matrix, and weight percent, mole percent and Schulz-Flory dis-
tributions, selectivity and conversion fractions, and C and H mate-
rial balances are calculated. The Tektronix graphics routines pro-
vide immediate plots of the hydrocarbon weight and Schulz-Flory
product distributions.
134 J. V. BAUER ET AL.

J C,/C,

,. I


m I
c: I
t C,-C ••



,.-- - - - - - ------------; I


- -- - DATA FLOW

Figure 2. Product analysis and data acquisition scheme

The total analysis of the Fischer-Tropsch product routinely

provides material balances to within 1-3%.

Catalyst Preparation

With the objective of max1m1z1ng selectivity and activity for

transportation fuels range hydrocarbons, precipitated, exchanged ,and
supported "conventional" catalysts, and novel, supported molecular
cluster catalysts have been prepared and studied in the slurry reac-
tor systems. Coprecipitated catalysts are prepared at constant pH
and temperature in a flow-through, tubular, stirred precipitation
reactor of the type described by Kolbel [3] and Deckwer [5].

Prior to slurry phase testing, catalysts are initially subjected

to a parametric screening program using a gas/solid tubular reactor~
to optimize catalyst preparation and activation procedures.


Sintered Fe203 - Baseline Slurry Catalyst

Initial slurry tests were made with an ammonia synthesis cata-

lyst, sintered Fe203 promoted with 2% A1203, 0.5% K20 and 0.7% CaO,
to provide baseline information. Prior to use, the catalyst was
ground to <200 mesh and reduced with H2 at 1 atm, 450°C and a GHSV
of 380 hr- l for 72 hr in a gas/solid tubular reactor. The reduced,
pyrophoric catalyst was then slurried under N2 in a deoxygenated
C25-C35 oil [15], and contacted with 0.5, 1.4 and 2.8:1 CO/H2 at -3.2
MPa, stir speeds of 800-1600 min-I, space velocities of 150-300 hr- l ,
and temperatures of 220-300°C.

Selected results, obtained over a continuous operating period

of 311 to 646 hr, are listed in Tables 1 and 2. Hydrocarbon weight
fractions of n-alkanes, 1-alkenes and branched isomers are plotted
against carbon number for CO/H2 ratios of 0.5, 1.4 and 2.8 in Fig-
ures 3-5.

As Table 1 shows, CH4 yields decreased and the hydrocarbon prod-

uct moved to increasing molecular weight as the CO/H2 ratio was in-
creased. At CO/H2 = 2.8, the CH4 yield was a low 4.7 wt%, but the
catalyst showed a decreased activity. A small trend to higher mo-
lecular weight product was also observed with increasing temperature
at a constant CO/H2 ratio. However, as Table 2 indicates, the rel-
ative amounts of n-alkanes, l-alkenes and branched isomers in the
total product varied little with temperature for CO-rich synthesis
gas, and were mainly dependent on the CO/H2 ratio. The largest 1-
alkene content occurred at CO/H2 = 1.4.

The test again confirmed the ability of the slurry reactor

system to operate continuously and efficiently with CO-rich synthesis
gas, as pointed out by Kolbel and others [3,5,6]. The water-gas
shift activity of the baseline catalyst increased with temperature,
and at 280°C, feed and usage ratios were equivalent for the 1.4 CO/H2
synthesis gas. The increase in higher molecular weight product
fractions with CO-rich synthesis gas was reflected in an increasing
contribution to the overall product analysis of hydrocarbons removed
directly from the slurry oil. No problems were experienced in fil-
tering this material directly from the reactor in a semi-continuous

Table 1. Summary of Selected Slurry Phase, Fischer-Tropsch Results

P, T, GHSV Activity, Selectivity, wt%

Catalyst hr-1 CO/H2
MPa °c mol syngas/kg/hr C1 C5-CU C9-C25 C!6

Fe203 3.31 280 250 0.5 26.2 17.4 37.6 17 .3 0.1

(baseline) 3.17 248 146 1.4 10.2 7.4 39.9 30.3 1.7
3.14 277 145 1.4 20.8 6.8 41.1 36.2 2.7
3.10 281 183 2.8 14.3 4.7 23.8 39.9 12.7
A1 2.03 252 318 0.9 9.9 9.5 41.3 50.8 0.1
A2 2.21 237 307 1.0 13.9 6.2 9.8 46.1 26.8 a
1.12 241 297 1.0 11.1 4.3 23.7 44.5 23.4
B 2.07 247 325 1.0 44.9 10.7 38.8 44.7 7.8
2.08 250 348 2.0 30.6 6.5 28.9 67.3 5.4

a64.0% C18-C35 !-
Table 2. The Dependence of a and Product Type on CO/H2 Ratio; Slurry Phase, Baseline Fe203a en"
Wt % of H~drocarbon Oxygenates, .:...
Slurry cone., P, T, Conv. , ;0
CO/H 2 a branched mol % of
wt% MPa °c mol % n-a1kanes 1-a1kenes o."
isomers product C/l
19.3 3.34 260 0.5 30.4 0.72 46.1 34.4 19.5 0.2 o
19.3 3.31 280 0.5 43.4 0.71 41.6 46.4 12.1 0.1 n
14.7 3.17 248 1.4 24.7 0.78 25.4 50.1 24.5 7.1 C/l
14.7 3.14 277 1.4 50.8 0.78 24.9 49.0 26.1 1.2
14.7 3.17 251 2.8 9.0 0.70, 40.7 40.5 19.1 2.2
0.93 b
14.7 3.10 281 2.8 27.7 0.70[, 40.0 44.2 15.8 2.8

aMean material balance for tabulated data: C 98.9 ± 1.5% and H 98.6 ± 2.1%
bFor >C10 hydrocarbons

138 J. V. BAUER ET AL.

COIH, - 0.5
T - 280 ' C
P - 3.31 MPa


5 10 15 20

Figure 3. Hydrocarbon weight distribution: FeZ03

Selective Catalysts AI, AZ and B [16]

Table 1 also lists some results using selective slurry catalysts
which were developed in the research program.

The hydrocarbon weight distribution obtained from catalyst Al

is shown in Figure 6. At O.g CO/HZ and Z.03 MPa, the slurry test
gave 50.8% C9-CZ5 and negligible CZ6 +' This is a considerably great-
er selectivity to C9-CZ5 than was obtained with the baseline catalyst
(30.3%) at a higher CO partial pressure (3.17 MPa and 1.4 CO/HZ),
which was shown to favor higher molecular weight product. Figure 6
shows that the higher C9-CZ5 fraction resulted from an enhancement

CO/H, - 1.4
T - 277' C
P - 3.14 MPa

Figure 4. Hydrocarbon
10 weight distribution:
Fe2 03

5 10 15 20 25 30

COtH, - 2.8
T - 281 "C
P - 3.10 MPa


5 10 15 20 25 30



Figure 5. Hydrocarbon weight distribution: Fe203

of hydrocarbon product in the CIO-Cll region.

A closely related catalyst A2, differing in minor details of

preparation, gave low yields of CH4 under slurry phase conditions
with the major part of the product occurring in the ClS-C3S region.
With 1.0 CO/H2 at 2.21 MPa, 64.0% of the hydrocarbon product was
found in this region, peaking at -C23, Figure 7, necessitating a com-
paratively rapid rate of sidestream withdrawal directly from the re-
actor. This product selectivity was found to be pressure dependent.
At 1.12 MPa, the hydrocarbon product contained a higher yield of CS-
Cll material, a low CR4 of 4.3%, but still retained a significant

COtH, - 0.9
T - 252°C
20 P - 2.03 MPa


5 10 15 20
Figure 6. Hydrocarbon weight distribution: Catalyst Al
140 J. V. BAUER ET AL.


COIH, - 1.0
T - 237 · C
20 P - 2.21 MPa






Figure 7. Hydrocarbon weight distribution: Catalyst A2

C18-C35 fraction, Figure 8. Even at the lower operating pressure,

catalyst A2 was significantly more selective for C9+ hydrocarbons
that. the baseline was at >3.3 times the CO partial pressure. During
this slurry test, feed and usage ratios were in close agreement, and
no catalyst deactivation was observed over a period of 457 hr.

Work is underway in our laboratories to better define the crit-

ical factors that determine the carbon number range of selectivity
with catalysts of the Al and A2 type.

25 r------------------------------------,
COIH, - 1.0
T - 241 · C
P - 1.12 MPa


5 10 15 20 25 30 35 40

Figure 8. Hydrocarbon weight distribution: Catalyst A2


COIH, - 1.0
T - 247"C
20 P - 2.07 MPa


5 10 15 20 25 30 35 40



Figure 9. Hydrocarbon weight distribution: Catalyst B

Catalyst B is another example of a slurry catalyst showing again

a selectivity dependence on the CO/H2 ratio, but a higher activity.
With 1.0 CO/H2 at 2.07 MFa, 44.7% of the product was found in the
C9-C25 region, but with a comparatively high CH4 yield of 10.7 wt%,
Figure 9. When the CO/H2 ratio was increased to 2.0, however, the
CH4 yield decreased to 6.5%, while the C9-C25 selectivity rose to
67.3 wt%, without a noticeable increase in higher molecular weight
C26+ materials, Table 1 and Figure 10. The concentration of hydro-
carbon product into a desirable fuels region with increasing CO/H2
ratio, clearly surpasses any effect observed with the baseline cata-

25 r----------------------------------,
COIH, - 2.0
T - 2SO' C
20 P - 2.08 MPa


5 10 15 20 25 30 35 40

Figure 10. Hydrocarbon weight distribution: Catalyst B

142 J. V. BAUER ET AL.

lyst, and with higher activity at lower CO partial pressures.


The Baseline Slurry Tests and the Schulz-Flory Distribution

The slurry tests of the baseline sintered Fe203 catalyst ful-

filled three functions. The first was to thoroughly test the analyt-
ical and data handline systems. The need to quantitate and analyze
all parts of the Fischer-Tropsch product, including heavier molecular
weight hydrocarbons that build up in the slurry reactor, in order to
obtain good carbon/hydrogen material balances, has already been em-
phasized. The second function was to determine the mass transfer
iimitations of the stirred, three phase test reactor. This function
is important in any experimental slurry catalyst testing system, be-
cause the effects of gas to liquid diffusional limitations must be
avoided or subtracted from the observed results to obtain meaningful
and reproducible kinetic parameters. To study mass transfer effects,
space velocity and stir speed were varied as a function of tempera-
ture. With CO/H2 > 1, and T ~ Z80°C, evidence was observed for sig-
nificant mass transfer limitations of the observed overall conver-
sion. Using the baseline results, a model of the stirred reactor
system has been developed to deconvolute the effects of mass transfer
and kinetic resistances and obtain activation energies and kinetic
parameters [8].

The third function was to obtain baseline activity and selectiv-

ity data in a slurry phase system. The Fischer-Tropsch synthesis is
a polymerization process, probably of surface adsorbed CHx species,
to produce higher hydrocarbon products. The growing chain can be
terminated either by desorption with or without hydrogenation, lead-
ing to alkanes or l-alkenes, or by reaction to give oxygenated spe-
cies. If the probability of incorporation of an additional surface
C atom into the growing chain is independent of the chain length, a
conventional Schulz-Flory distribution of products occurs. This is
described in detail in the work referred to in reference [9]. In
logarithmic form, this type of product distribution can be repre-
sented by the following equation:


where Wi is the weight fraction of the product with carbon number

Ci' and u is the probability of chain growth, with values of 0 + 1.
If this type of distribution is followed, a plot of 10g(Wi/Ci)
against carbon number Ci produces a straight line of negative slope,
log u.

Because of the unselective nature of this type of polymerization

process, the Schulz-Flory distribution of products, if adhered to,
has significant implications for the maximum yield of transportation


C.. +
C 60


0.4 0.6 0.8 - - 1 . 0


Figure 11. Variation of hydrocarbon fraction with u: Schulz-Flory


Table 3. Baseline and Schulz-Flory Product Fractions, wt%

C9- C25

Schulz-Flory Maxima
gasoline range 0.76 5.8 47.6 31.8 0.7
diesel range 0.88 1.4 31.9 S4.l 12.9

Baseline Fe203
CO/H2 O.S 0.70 17.4 37.6 17.3 0.1
CO/H2 = 1.4 0.78 6.8 41.1 36.2 2.7

fuel product fractions. Figure 11 shows the variation, in weight

percent of the total product, of the gasoline range hydrocarbons CS-
C11, and the diesel range hydrocarbons C9-C2S' with changes in the
chain growth probability u. As illustrated in Figure 11, the Schu1z-
Flory distribution imposes limits on the maximum amount of these
product fractions of 47.6 wt% CS-C11 and S4.1 wt% C9-C2S. These
values are listed in Table 3, along with the necessary yields of
methane and C26+ hydrocarbons imposed by the unse1ective statistical
144 J. V. BAUER ET AL.

Curve CO/H. MPa a
-1 .. (a) 3.10 281 0.70,0.93
(b) 1.4 3.14 277 0.78
(e) 0.5 3.31 280 0.71
-2 {
W· \ ......
LOG (_I)
Cj ........ , ° 0 • ---.

-3 ._.-' ':"'7: - .. (a) .,_._,_._._.

• (b)~ ..... ,

-4 • (e) •


5 10 15
____ ____ ____




Figure 12. Hydrocarbon Schulz-Flory distribution: Fe2D3

Many reviews have indicated that most Fischer-Tropsch products

adhere to the Schulz-Flory distribution, irrespective of the catalyst
type [6,10]. Satterfield et al. concluded that the Schulz-Flory dis-
tribution generally held for iron-based catalysts with values of a
ranging from 0.55 to 0.94 [6]. In a slurry test of a similar sin-
tered Fe2D3 ammonia synthesis catalyst, with a CD/H2 ratio of 1.5 and
a pressure of .79 MPa at 269°C, they observed a close fit of the
products from the slurry reactor in the range Cl-C2l to a Schulz-
Flory distribution with a value of a = 0.7 [6].

Table 3 compares results from the baseline catalyst test report-

ed in this paper with the Schulz-Flory maximum product fractions. It
can be seen that for CO/H2 = 1.4, the product fractions are closest
to the gasoline range Schulz-Flory maxima with a higher amount of
C26+ material.

Table 2 shows that the values of a obtained from the slurry

baseline Fe203 catalyst are independent of temperature in the range
250-280°C, but depend strongly on the CO/H2 ratio. In Figure 12, the
hydrocarbon product distributions of Figures 3, 4, and 5 are plotted
on a Schulz-Flory basis. For H2-rich synthesis gas, i.e. CO/H2 =
0.5, the hydrocarbon product very closely follows a straight line
Schulz-Flory distribution with a value of a = 0.71 over the range Cl-
C23. For CO-rich synthesis gas with CO/H2 = 1.4, the product distri-
bution still closely approximates a straight line Schulz-Flory dis-
tribution, at least up to C23, but with a higher value of a = 0.78,
reflecting the shift of the overall product to higher molecular
weights (Table 1). However, for the highest CO ratio synthesis gas,
CD/H2 = 2.8, the hydrocarbon distribution plotted on a Schulz-Flory

basis falls into two distinct regions separated at -ClO, with the
ClO-C30 hydrocarbons following a separate line with a high a value
of 0.93.

The phenomenon of two a values for the Schulz-Flory distribution

observed with CO-rich synthesis gas, may be due to the participation
of olefinic products in a secondary polymerization process. Although
reincorporation of olefinic primary products with a probability in-
dependent of carbon number leads to another Schulz-Flory distribution
with a higher value of a [11], this process may become more selective
in the slurry phase. The relative involatility of higher olefins
and a diffusional cage effect around the catalyst particles in the
slurry oil may cause a preferential reincorporation above a partic-
ular carbon number. This would be consistant with the observation
of the effect at high CO/H2 ratios, where the primary product is ex-
pected to be olefinic, although as Figure 5 shows, much of the higher
molecular weight product is hydrogenated through the water-gas shift
reaction. Further experimental work is necessary to confirm this.

Selective Slurry Catalysts

Table 4 lists some published data for (i) the commercial Fisch-
er-Tropsch units at SASOL [10,13], (ii) a pilot plant three-phase
bubble column reactor at Rheinpreussen, 1.6 m id x 8.6 m high [3],
and (iii) some bench scale slurry column experiments carried out by
KBlbel [12]. The operation at Rheinpreussen was geared to maximize
the yield of gasoline. Some doubt exists about the overall product
analysis, since the data reported in Table 4 was stated to be from
a "sample from large storage tanks which contained the products from
one of the longer operating periods" [3]. It may not, therefore,
fully represent the percentage of light products obtained from the
plant, which were removed by oil scrubbing and activated carbon ad-
sorption. With this in mind, a comparison of the product fractions
of Table 3 with the data of Table 4 indicates that the SASOL reactors
and previous slurry phase operations gave products that again were
unselective and of the Schulz-Flory limited type. Satterfield et
al. have also shown that the slurry phase data obtained by KBlbel can
be correlated by a Schulz-Flory distribution using the appropriate
value of a [6].

However, the selective slurry catalyst results reported here

show large deviations from a standard, straight line Schulz-Flory
distribution. In Figure 13, hydrocarbon product distributions ob-
tained with the selective slurry catalyst A2 are plotted on a log
(Wi/Ci) basis against carbon number, with a baseline catalyst distri-
bution for comparison. Figure 13 clearly shows the enhanced product
deviation from the limitations of the Schulz-Flory distribution ob-
served with this catalyst at 2.21 MPa for the C18-C35 region. The
effect of pressure in obtaining this product enhancement is shown by
a comparison of curves (a) and (b).

Table 4. A Comparison of Commercial SASOL and Prior Slurry Phase Fischer-Tropsch Data

SASOL [lOaD] Rheinpreussen KBlbel Bench Scale Slurry [12]

Arge Synthol
Reaction slurry
fixed entrained Low Medium High
1.6 id x 8.6 m
bed solid MW MW MW

CO/H2 0.6 0.5 1.5 1.5

T, °c 220-250 320 268 268
P, MPa 2.6 2.0 1.2 1.1
Conversion, % 70 85 89 88
Recycle 2.5 2 0 0
Space Time Yield 50-70 90 41 32
(kg/m3 /h)
Cl, wt% 5 10 6.8 (Cl+C 2) a 6.3 (Cl+C2)
Approx. Gasoline
Range, Wt% 22 39 (C5-C12) 53.6 (40-l80°C) 68.1 40.0 7.1 (C5-l90°C) b
Approx. Diesel
Range, Wt% 21 6 (C13-C21) 10-12 (180-320°C) 13.3 44.0 41.2 (190-450°C)
Higher Hydro-
carbons, Wt% 35 5 (C22+) 1. 9 (>320 0 C) 0.6 9.1 49.5 (>450°C) ~
aStorage tank sample c
bwt% of C3+ products :IJ

o r-----------------------------------------~
Curve COIH. MPa ~ Catalyst
-1 (a) 1.0 2.21 237 } A2
\ .... (b) 1.0 1.12 241
~ (e) 1.4 3.17 248 Fe 2 0 3
-2 .~.I\ •••••
W· , '.
LOG (i-) ~:~,
.' ....... (a)~
-3 .........-'"'_"""'::-... ' - ' - ' (b) -.
. • (e)·· . '.,
...... .........
-4 '""

-5L-__ ~ __ ~ __ ~ ____ ~ __ ~ __ ~ ____ ~ __ ~

5 10 15 20 25 30 35 40


Figure 13. Hydrocarbon Schulz-Flory distribution: Slurry Catalyst A2

Product distributions obtained from the slurry test of catalyst

B are plotted on a log(Wi/Ci) against carbon number basis in Figure
14. A baseline product distribution obtained at a higher CO partial
pressure, which should enhance the yield of higher molecular weight
product, is also plotted for comparison. The large deviation from
the standard Schulz-Flory distribution observed with CO/H2 = 2.0 in
the diesel range region C9-C25 can be clearly seen in this figure.
The 67.3 wt% product obtained in the C9-C25 region represents a 25%

Curve CO/H. MPa °C Catalyst
(a) 2.0 2.08 250
-1 (b) 1.0 2.09 247 J B
(e) 2.7 3.17 253 Fe.O.


LOG( Cjl )



-5U-__ ~ __ ~ __ ~ ____ ~ __ ~ __ ~ ____ ~ __ ~

5 10 15 20 25 30 35 40

Figure 14. Hydrocarbon Schulz-Flory distribution: Catalyst B

148 J. V. BAUER ET AL.

increase over what was previously thought to be the limit of 54.1%

imposed by Schulz-Flory distribution.

The enhanced selectivity, and the breakdown of the Schulz-Flory

distribution observed with these catalysts, may reflect a variation
in the chain growth probability a with the chain length Ci. Nijs
and Jacobs [14] have proposed a metal particle size effect on the
product distribution that would be observable with uniformly dis-
persed catalysts under process conditions tending to give high a



We have demonstrated that novel catalysts operated in the slurry

mode can overcome the limitations on the yield of fuel product frac-
tions that were previously thought to be imposed by the standard
Schulz-Flory distribution. In particular, improved selectivity has
been obtained for the C9-C25 diesel fuel region. This is particular-
ly important since the Fischer-Tropsch product, with its potential
for high n-alkane yield, appears to be well suited for the production
of diesel fuel. With a particular slurry catalyst, selectivity is
generally a function of CO partial pressure, involving the operating
pressure or the CO/H2 ratio.

Slurry Phase Reactors

The slurry phase Fischer-Tropsch reactor is ideally suited for

producing enhanced yields of material in the diesel fuel range with-
out encountering conditions, such as plugging due to wax build-up,
that would make the use of fixed-bed or gas entrained solid reactors
less suitable for the higher molecular weight hydrocarbon range.
High product yields in this range can be successfully accommodated
without any interruption in operation, and with the use of CO-rich
synthesis gas. Any higher molecular weight product build-up that
does occur can be withdrawn as a sidestream from the reactor. Cata-
lysts with water-gas shift activity allow CO-rich synthesis gas feed
and usage ratios to be equalized. The slurry reactor, therefore, im-
proves the integration of indirect liquefaction with newer generation
coal gasifiers, by allowing the continuous production in a single-
stage of a fuel range product that requires minimal further process-

Future Research and Development

As part of our continuing research and development program into

the slurry phase Fischer-Tropsch process, further work is being car-
ried out on increasing the activity and selectivity of slurry cata-

lysts for the fuel range hydrocarbons, particularly C9-C25. Ongoing

cold flow modelling studies are defining the hydrodynamics of three-
phase bubble columns, with the objective of maximizing the space
time yield of this type of reactor [lb).


The authors wish to acknowledge the support of this work by the

U. S. Department of Energy (Contract Number DE-AC22-80PC3002l) and
by Air Products and Chemicals, Inc.


1. a. P. N. Dyer, R. Pierantozzi, B. W. Brian, and J. V. Bauer,

Quarterly Technical Progress Reports 1-8, US DOE PC/3002l,
b. B. W. Brian and P. N. Dyer, Abstr. No. INDE-13, l85th ACS
National Meeting, Seattle (March 20-25, 1983).
2. H. H. Storch, N. Golumbic, and R. B. Anderson, "The Fischer-
Tropsch and Related Syntheses", Wiley, New York (1951).
3. H. K6lbel and M. Ralek, Catal. Rev. Sci. Eng., 21:225 (1980).
4. M. L. Pout sma , "Assessment of Advances Process Concepts for the
Liquefaction of Low H2/CO Ratio Synthesis Gas," ORNL-5635
5. W. D. Deckwer, Y. Serpemen, M. Ralek, and B. Schmidt, Ind. Eng.
Chem., Proc. Des. Dev., 21:222 and 21:231 (1982).
6. C. N. Satterfield, G. A. Huff, and J. P. Longwell, Ind. Eng.
Chem., Proc. Des. Dev., 21:465 (1982).
7. J. V. Bauer and P. N. Dyer, Chem. Eng. Prog., 78(9):51 (1982).
8. P. N. Dyer, J. V. Bauer, B. W. Brian, and R. L. Parsons, to be
9. a. R. B. Anderson, "Catalysis", Reinhold, Vol. IV, 29 (1956).
b. P. J. Flory, "Principles of Polymer Chemistry", Cornell
University Press, 318 (1967).
c. G. Henrici-Olive and S. Olive, Agnew. Chem., Int. Edn., 15:
136 (1976).
d. M. E. Dry, Ind. Eng. Chem., Prod. Res. Dev., 15:282 (1976).
10. L. Caldwell, "Selectivity in Fischer-Tropsch Synthesis," CSIR
Report CENG330, Pretoria, S. Africa (1980).
11. S. Novak, R. J. Madon, and H. Suhl, J. Catal., 77:141 (1982).
12. H. Kolbel, P. Ackerman, and F. Engelhardt, Erdol und Kohle,
9:225 (1956).
13. M. E. Dry, CHEMTECH, 744 (Dec. 1982).
14. H. H. Nijs and P. A. Jacobs, J. Catal., 65:328 (1980).
15. 0-120 Paraffin Oil, Fisher Scientific Co., Fairlawn, NJ 07410
16. Details of these catalysts will be published separately.


v. U. S. Rao,* R. J. Gormley, R. R. Schehl, K. H. Rhee,

R. D. H. Chi and G. Pantages

U. S. Department of Energy
Pittsburgh Energy Technology Center
P. O. Box 10940
Pittsburgh, PA 15236


It is well known that the medium-pore zeolite ZSM-5 can cata-

lytically convert methanol or other oxygenates to hydrocarbons con-
taining aromatics [1]. The conversion is normally carried out in
the temperature range 300°-400°C. The acid function of ZSM-5 is
responsible for the catalytic synthesis of aromatics. The shape-
selective behavior stemming from the medium-pore size (diameter of
5.5 X results in a sharp cut-off in the aromatic product distribu-
tion at CIO' The reaction path [2-4] involves the following steps:
(a) dehydration of methanol to dimethylether and subsequently to
ethylene and propylene, (b) oligomerization of light olefins to
heavier olefins, and (c) isomerization, cyclization, and aromatiza-
tion. The liquid hydrocarbon product contains blanched paraffins,
blanched olefins, naphthenes, and aromatics, and.has a high octane

If light olefins are used as feed, ZSM-5 can catalyze the con-
version to a high octane liquid product via steps (b) and (c) out-
lined above. Since olefins and oxygenates are obtained as some of
the products of synthesis gas conversion by the Fischer-Tropsch
reaction, there is considerable interest in combining a typical
Fischer-Tropsch metal such as Fe, Co, or Ru with ZSM-5 to form a bi-
functional catalyst. It has been shown that such a catalyst can
convert synthesis gas to gasoline-range hydrocarbons in one step

152 v. U. S. RAO ET AL.

The activity and selectivity of metal-zeolite catalysts in

the conversion of synthesis gas can depend on the method of cata-
lyst preparation. Some aspects requiring experimental study are
the following:

(a) state of dispersion of the metal,

(b) influence of the metal-containing component on the cata-
lytic function of the zeolite, and
(c) possible deactivation of the metal and zeolite components
during synthesis gas conversion.

In the present study, attempts were made to examine the behavior of

cobalt-ZSM-S catalysts in the conversion of synthesis gas. The cat-
alysts prepared by solution impregnation methods were examined, and
it was realized that the method could result in partial ion exchange
of the acidic protons in ZSM-S by cobalt ions. Additional catalysts
were prepared by the physical admixture of precipitated cobalt oxide
with ZSM-S. The catalytic activities of the catalysts prepared by
the two methods will be compared with each other. Ethylene conver-
sion studies were performed on H-ZSM-S and Co-ZSM-S to examine the
influence of metal addition on the acid-catalyzed zeolite function.

The catalysts were characterized using infrared spectroscopy

to examine the acid sites prior to and after addition of the metal
component to the zeolite. Ammonia adsorption studies were performed
to obtain the strength of the acid sites. Hydrogen chemisorption
studies have been found to be useful in elucidating the state of the
metal component. The results from the characterization will be dis-
cussed along with the catalytic activity results in an effort to
understand the different behavior of the catalysts prepared by the
two methods.


Preparation of Catalysts

The ZSM-S, with Si02/A1203 - 30, was synthesized using the

procedure given in the patent literature by Argauer and Landolt [9]
with minor modifications. The zeolite was calcined in air at S38°C
to decompose the tetrapropylammonium cation (TPA+). It was con-
verted to the ammonium form by three successive exchanges using
NH4CI solution.

In the solution-impregnation method, the aqueous cobalt(II)

nitrate solution was added to NH4-ZSM-S until incipient wetness was
reached. The atmosphere surrounding the mixture was evacuated after
the solution was added. The mixture was dried with stirring at
10QoC, and the material was pelleted into 3 mm diameter tablets.

In the physical admixture method, the metal oxide was initial-

ly made as a precipitate by adding a 10 wt% sodium carbonate solu-
tion at 70°C to a 10 wt% cobalt nitrate solution also at 70°C until
a pH of 7 was reached. Evolution of C02 was observed during the
mixing of the solutions. The precipitate was filtered, washed with
water thoroughly to ensure a content of less than 0.1 wt% Na, and
dried at 110°C. The dry metal oxide and the dry zeolite were each
sieved to -200 mesh, then thoroughly mixed and pelleted.

The catalyst pellets were calcined in air at 450°C for one

hour to convert the NH4-ZSM-5 to the H-form. Parallel studies using
infrared spectroscopy showed deammoniation was nearly complete under
these conditions.

Synthesis Gas Conversion Tests

The catalyst sample' weighing 1.35 g was introduced into a ver-

tical, down-flow, reactor tube of 1 cm inner diameter. The bed
height of the catalyst was 2.5 cm. The catalyst was heated in a
nitrogen atmosphere at 21 bar to 450°C and kept at that temperature
for 16 hr. The temperature was lowered to 190°C, and hydrogen at
21 bar and a flow rate of 70 ml/min was introduced. The temperature
of the catalyst was raised to 350°C at the rate of 2°C/min and the
catalyst was kept at that temperature for 16 hr. The temperature
was lowered to 250°C, and the pressure to 7 bar. Synthesis gas with
the composition H2/CO = 1 was introduced and the pressure was slowly
raised to 21 bar. The temperature was raised to 280°C and a flow
of 0.659 gig cat/hr was maintained.

The products of the reaction were mainly C02, H20, and hydro-
carbons. A hot trap at 175°C, following the reactor, collected wax
and heavy oil. A cold trap at O°C, following the hot trap, collect-
ed water, gasoline-range products, and light oil. The remaining
product that was not in the condensed phase and the unconverted feed
gas moved past the back pressure regulator. At atmospheric pressure,
the gas mixture passed through a wet test meter that measured the
total volume of gas that had passed during a test period. Samples
of the exiting gas were collected periodically and analyzed using a
gas chromatograph for H2' CO, C02' and Cl-C7 hydrocarbons. The
liquid product collected in the cold trap was separated physically
into aqueous and hydrocarbon fractions. A mass balance was performed
to provide information such as that shown in Table 1. It is assumed
that the liquid hydrocarbon product consists of Cs+. Material re-
covery, as shown by the mass balance procedure, was usually better
than 96%.

The product collected in the hot trap was found to be very small
in quantity in comparison to that collected in the cold trap. The
hydrocarbon product from the latter was analyzed by ASTM-D-2887 GC
simulated distillation. It was also separated into aromatics, ole-
154 v. U. S. RAO ET AL.
Table 1. Conversion and product distribution from Co-ZSM-5 cata-
lysts during the initial 24 hour period. H2/CO = 1;
P = 21 bar; T = 280°C; Feed rate = 0.659 gig cat/hr

Physical admixture
Method of Aqueous with precipitated
adding cobalt nitrate solution cobalt oxide

Co in catalyst (wt%) 2.7 5.9 9.0 2.8 5.4 8.5

CO conversion % 21.0 54.6 56.5 24.7 38.2 55.0
H2 conversion % 34.3 80.8 85.8 46.0 59.4 66.5
CO + H2 conversion % 27.4 67.3 70.7 35.3 48.5 60.8

Product composition (wt%)

C02 8.1 13.2 18.6 10.3 10.2 12.0
H2 0 56.6 50.9 46.1 53.7 52.7 51.3
CMu 35.3 35.8 35.2 35.9 37.0 36.6

Composition of CHn (wt%)

CH4 33.6 24.0 24.4 32.5 28.3 31.2
C2 H4 0.0 0.4 0.0 0.0 0.0 0.0
C2 H6 5.2 2.3 2.7 3.4 3.5 3.9
C3 H6 0.0 0.6 0.8 0.0 0.0 0.0
C3 H8 7.7 2.7 2.7 12.5 9.6 7.1
C4 H8 0.9 1.4 0.8 0.0 0.0 0.0
C4 HlO 18.7 6.6 5.4 23.0 22.3 15.7
C5+ (exc1. wax) 33.8 60.8 62.2 28.6 36.3 41.1
Wax 0.0 1.1 0.8 0.0 0.0 0.9

Liquid product composition (vol%)

Aromatics 23.0 19.5 65.5 48.5
Olefins 30.5 32.5 0.5 0.5
Saturates 46.5 48.0 34.0 51.0

fins. and saturates by ASTM-D-1319 column chromatography with fluo-

rescence indicator adsorption (FIA). The results of such separation
are included in Table 1. The combined aromatic and paraffin frac-
tions were analyzed by ASTM-D-2789 mass spectrometry to find the
percentage of cycloparaffins. and the distribution of the a1ky1ben-
zenes by carbon number up to C12'

Ethylene Conversion Tests

The catalyst sample (H-ZSM-5 or Co-ZSM-5). weighing 1.5 g. was

introduced into the reactor. Hydrogen at 1 atm was passed over the
catalyst at 350°C for 16 hr. The temperature was lowered to 320°C.

and the reaction mixture, consisting of 30 vol% C2H4 and 70 vol% H2,
was introduced at 1 atm and a flow rate of 0.983 gig cat/hr. The
liquid product, which consists of hydrocarbons, was collected in a
trap at O°C. The remaining effluent, which was in the gas phase,
consisted of H2 and hydrocarbons. Samples of the effluent were ana-
lyzed for H2 and Cl-C7 hydrocarbons. Mass balances were performed
to obtain the information shown in Table 2. The liquid hydrocarbon
product was analyzed as described earlier.


Catalytic Conversion of Synthesis Gas

The experiments were designed to provide a comparison of the

selectivities of metal-zeolite catalysts prepared by (i) the impreg-
nation of ZSM-5 with cobalt nitrate solution and (ii) the admixture
of precipitated cobalt oxide with ZSM-5. Three metal loadings in
the range 2.7 to 9.0 wt % cobalt in the catalyst were examined for
each method of preparation. The results are given in Table 1. The
change in CO and H2 conversion with metal loading is shown in
Figure 1. The following features are worthy of note:

(i) In the case of the solution-impregnated catalysts, the

CO + H2 conversion increases rather slowly with cobalt concentration
in the range 5-9% cobalt. The increase is more pronounced in the
range of 2-5% cobalt. In the case of the physically admixed cata-
lysts, the variation of CO + H2 conversion with cobalt loading is
nearly linear.

(ii) At the lowest cobalt concentration studied, a higher frac-

tion of gas phase (Cl-C4) hydrocarbons is produced than with cata-
lysts having higher cobalt concentrations, especially in the case
of the solution-impregnated catalysts.

(iii) In the case of the solution-impregnated catalysts, the

liquid hydrocarbon yield, represented approximately by the C5+ prod-
uct, increases appreciably when the cobalt loading is increased from
2.7% to 5.9% and remains nearly steady when the cobalt loading is
increased from 5.9% to 9.0%. In the case of the physically admixed
catalysts, there is a more gradual increase in the C5+ fraction in
the product hydrocarbons as the cobalt loading is increased from
2.8% to 8.5%. It should be mentioned that in all cases, nearly 90%
of the liquid hydrocarbon was in the gasoline boiling range as deter-
mined by simulated distillation.

(iv) The physically admixed catalysts provide a much higher

fraction of aromatics in the liquid hydrocarbon product than the
solution-impregnated catalysts. The percentage of olefins in the
product from the physically admixed catalysts is nearly zero, indi-
156 V. U. S. RAO ET AL.

Table 2. Conversion of 30 vol% C2H4 + 70 vol% H2 mixture over

H-ZSM-5 and Co-ZSM-5. Si02/A1203 = 23. Feed rate =
0.983 gIg cat/hr; Temperature = 320°C

Catalxst H-ZSM-5 1.7% Co-ZSM-5 1.7% Co-ZSM-5
Physically Ion-Exchanged

Period (hr) 0-4 4-8 0-4 4-8 0-4 4-8

C2H4 conversion % 91.8 83.9 96.1 91.7 70.5 60.2
H2 conversion % 8.5 8.0 11.7 12.5 10.5 10.3

Product comEosition ~wt%)

CH4 0.0 0.0 0.5 0.6 0.0 0.0
C2 H6 8.6 10.8 16.8 20.9 21.1 24.6
C3 H6 5.9 8.4 3.6 5.2 10.8 12.4
C3 8 7.9 5.7 8.7 6.1 4.1 2.8
C4 H8 7.2 10.5 5.2 7.4 12.7 13.9
n- C4HlO 7.2 6.7 7.4 6.3 6.3 6.4
i-C4HlO 11.5 8.2 10.6 7.7 4.9 3.7
C5+ 51.7 49.7 47.1 46.0 40.0 36.1

Liguid Eroduct comEosition

Aromatics 56.5 38.5 60.0 41.0 34.0
Olefins 24.5 48.0 19.0 42.0 55.0
Saturates 19.0 13.5 21.0 17.0 11.0

eating a very efficient conversion of olefinic intermediates to aro-

matics by these catalysts. More than 90% of. the aromatic product
consisted of alkylbenzenes, with C8 and C9 alkylbenzenes the leading
constituents. With the catalysts containing the lowest amount of
cobalt (-2.7%), the liquid product collected during a 24 hr period
was not sufficient for the FIA analysis, which requires about 1 ml
of sample.

The solution-impregnated catalysts with 5.9% and 9.0% Co yielded

gasoline-range selectivity of nearly 57%, higher than the amount
predicted (48%) by Schulz-Flory polymerization kinetics [10]. NMR
studies showed a high degree of branching of the olefin and saturate
fractions in the liquid product obtained from such catalysts. In
combination with the aromatic fraction, which is nearly 20 vo1%, the
liquid hydrocarbon product is of the high octane gasoline type.

Some catalysts were examined at a reaction temperature of 300°C

with the other conditions indicated in Table 1. At the higher tem-
perature, the conversion was higher. The methane yield was substan-



...= 60
ffi 50

! 40 Catalysts: Cabalt-ZSM-5
o • Salution impregnated
• Physically admixed
30 Feed: H2/CO • 1.0
Test canditlons: Ts 280°C, P.21
bar, WHSV-0.66
Periad: 0-24 haur.


Figure 1. Variation of synthesis gas conversion with percent

cobalt in Co-ZSM-5 catalysts. Run conditions are indi-
cated in Table 1.

tially higher, the C5+ yield was lower, and there was an increase
in the aromatic fraction of the liquid product. For example, the
solution-impregnated catalyst 5.9% Co-ZSM-5 of Table 1, when examined
at a reaction temperature of 300°C, showed CO conversion of 63.0%.
The methane selectivity was 42.0%, the C5+ selectivity was 38.7%,
and the fraction of aromatics in the liquid hydrocarbon product was

An experiment was performed to examine the change in conversion

and selectivity of a physically admixed 9.2% Co-ZSM-5 catalyst over
a period of 14 days. The results are shown in Figure 2. There is
a gradual decline with time on stream in (H2 + CO) conversion and
C5+ selectivity, while there is an increase in the selectivity to
methane. There is a decrease in percentage of aromatics and a corre-
sponding increase in percentage of olefins in the liquid product with
time, while the percentage of saturates in the liquid products re-
mains relatively steady. The deactivation of the metal component
results in decreasing conversion, and that of the zeolite component
is possibly responsible for the decrease in percentage of aromatics.

Catalytic Conversion of Ethylene

The conversion of ethylene over H-ZSM-5 and the metal-zeolite

samples was used as a method for comparison of their catalytic acid
158 v. U. S. RAO ET AL.

Catalyst· Cabalt-Z SM-5 (physically admixed)
containing 9.2 w t. % cobalt. -

• 60 ~ 60
~ A..... •••
~..J 40
~~---- ..----_.A.~--r-.=:.::::-:::..=-_
_ . - .... ..
_ • ...-. 40~
Feed: Hz/CO = I • Conversion
20 T=280°C, P=21 bar
WHSV=0.66 • C5+ 20 6u
• C,

o 14

° .,Jt.
> 60
Z "".
",,"" "" ""

e I
0 20 I
::i I • Aromatics
",II • Olefins

0 14

Figure 2. Variation of synthesis gas conversion and product selec-

tivity with time on stream for a 9.2% Co-ZSM-5 catalyst
prepared by the physical admixture method. Run conditions
as in Table 1.

function. The possibility of partial ion exchange of the acid pro-

tons by C02+ in the solution impregnated Co-ZSM-5 samples was consid-
ered. Repeated washing of a solution-impregnated 5.9% Co-ZSM-5 sample
resulted in a sample containing 1.7% Co. The cobalt remaining in the
sample after washing could not be reduced to metallic cobalt after ex-
posure to flowing hydrogen at 350°C, and the sample exhibited exceed-
ingly low catalytic activity for synthesis gas conversion. The ESCA
studies showed the presence of C02+. It was surmised that the 1.7%
Co-ZSM-5 sample contained cobalt mainly in ion-exchanged form. For
purposes of comparison, a sample of 1.7% Co-ZSM-5 was prepared by
the physical admixture method.

Ethylene conversion studies were performed at 320°C with a

C2H4/H2 = 30/70 vol% reactant gas mixture on the following samples:

(B) 1.7% Co-ZSM-S (physically admixed)
(C) 1.7% Co-ZSM-S (ion-exchanged).

The results are shown in Table 2. The data have been provided for
two separate periods, 0-4 and 4-8 hr on stream, respectively.

During the 0-4 hr period, samples A and B show ethylene conver-

sion rates much higher than that of sample C. Samples A and B also
show much higher selectivity for CS+ product and for aromatics. It
is indicated that the acid function of the zeolite is weakened in
sample C by the presence of cobalt in ion-exchanged form. The aro-
matics obtained from ethylene conversion were predominantly (>90%)
alkylbenzenes. The distribution of the alkylbenzenes was similar
to that obtained from synthesis gas conversion over metal-zeolite
catalysts. A comparison of the alkylbenzene products is shown in
Table 3. Mostly, C7-CIO alkylbenzenes are formed in both types of
reaction. The near parallelism in the product distribution from
synthesis gas and ethylene conversion suggests similar pathways to
aromatic compounds. In the case of synthesis gas conversion, the
light olefins are formed initially by the F-T component, namely
metallic cobalt.

The choice of the reaction temperature for synthesis gas con-

version over Co-ZSM-S catalysts is governed by considerations of F-T
reactions and the subsequent acid-catalyzed conversions by ZSM-S.
The conversions over ZSM-S are known to be facile in the temperature
range 300°-400°C. The F-T reactions over cobalt are usually con-
ducted in the temperature range 180°-260°C, if a substantial CS+
selectivity is desired. As a compromise, the reaction temperature
for most of our experiments on synthesis gas conversion over metal-
zeolite catalysts was chosen as 280°C. In retrospect, this temper-
ature appeared to be suitable for obtaining CS+ selectivity in excess
of 60% over some catalysts, with the liquid product containing suffi-
cient aromatics and branched hydrocarbons required for high octane

Some experiments were performed to examine the influence of

temperature on the conversion of ethylene over H-ZSM-S. Table 4
provides data on the conversion at 280°C and 320°C over a sample of
H-ZSM-S with Si02/A1203 = 33. The conversion of ethylene is lower
at 280°C in comparison to that at 320°C. The fraction of aromatics
is lower at a reaction temperature of 280°C in comparison to that
obtained at the higher temperature. However, the ethylene conversion
(52.6%) and the fraction of aromatics (29.5%) in the liquid hydro-
carbon product are sufficiently high to justify the testing of
metal-ZSM-S catalysts for synthesis gas conversion at 280°C.
160 V. U. S. RAO ET AL.

Table 3. Alkylbenzenes from synthesis gas

conversion and ethylene conversion

Carbon Synthesis gas conversiona Ethylene conversion b

number over 5.4% Co-ZSM-5 (vol%) over H-ZSM-5 (vol%)

6 0.0 0.0
7 11.6 8.4
8 37.9 31.0
9 30.8 36.2
10 14.5 19.0
11 3.8 4.4
12 1.3 1.0

aphysically admixed catalyst. Synthesis gas conversion at 280°C

shown in Table 1.
bCatalyst (A) and conditions (T = 320°C) same as in Table 2.

A comparison of the ethylene conversion data (Table 2) at 320°C

for the 0-4 hr period with those for the 4-8 hr period shows a modest
decrease in ethylene conversion with time on stream. More noticeable
is the decrease in the fraction of aromatics in the liquid product.
The fOrffiation of aromatics by the dehydrocyclization of C6+ olefins
is catalyzed [2,3] by the strong acid sites in H-ZSM-5. Deacti-
vation of the strong acid sites with tim~ on stream would lead to
reduction in the formation of aromatics and an increase in the liquid
phase olefins formed by the oligomerization of the light olefins.
The latter reaction is also catalyzed by the acid sites but can be
mediated by sites of lower acid strength than those required for the
formation of aromatics. Microcalorimetric studies [11] and ammonia
desorption studies [12] have shown a distribution of acid site
strength in H-ZSM-5. The heat of ammonia adsorption [13] varies
from 60 to 170 kJ mol- l for the acid sites. The strong acid sites
possibly deactivate more rapidly than those of intermediate strength,
owing to the ability of the former to synthesize aromatic coke

The decrease in percent aromatics in the product with time on

stream is less rapid (see Figure 2) in the synthesis gas conversion
reaction at 280°C than in the ethylene conversion reaction at 320°C
(Table 2). Among the factors that influence the deactivation rate
of ZSM-5 is the presence of water in the reaction mixture. It has
been shown [4] in C2H4 + H20 conversion over H-ZSM-5 that an increas-
ing amount of water in the feed slows down the rate of deactivation
of the catalyst. In the F-T reaction over Co-ZSM-5 catalysts (Table
1), a substantial amount of H20 is formed. The presence of water

Table 4. Conversion of 30 vol% C2H4 + 70 vol% H2 mix-

ture over H-ZSM-5 (Si02/A1203 = 33) at 280°C
and at 320°C. Feed rate = 0.983 gig cat/hr

Temperature 280°C 320°C

Period (hr) 0-4 0-5
C2H4 conversion % 52.6 97.3
H2 conversion % 11.6 15.3

Product composition (wt%)

CH4 0.0 0.0
C2 H6 27.9 21.8
C3H6 6.4 2.9
C3 H8 2.9 6.6
C4 H8 9.2 3.6
n- C4HlO 5.8 6.4
i-C4HlO 4.5 13.9
C5+ 43.3 44.7

Liquid product composition (vol%)

Aromatics 29.5 65.5
Olefins 53.0 12.0
Saturates 17 .5 22.5

in the reaction mixture may have been responsible for the slower
decline in the rate of aromatics production than would have been the
case otherwise.

Catalyst Characterization

The catalysts were characterized using hydrogen chemisorption,

magnetic studies, IR spectroscopy, X-ray photoelectron spectroscopy
(XPS), gravimetric ammonia adsorption, and X-ray diffraction line
broadening studies. The details of the investigations will appear
in separate publications. In this paper, results of characterization
studies that are relevant for the understanding of the catalytic
behavior will be mentioned briefly.

The chemisorption experiments [14] on the solution-impregnated

Co-ZSM-5 samples, after exposure to flowing hydrogen at 350°C, showed
hydrogen uptakes at 100°C of 0.0, 1.9, and 4.0 ml/g for samples with
2.7, 5.9, and 9.0 percent Co, respectively. The results indicate
that in the solution-impregnated 2.7% Co-ZSM-5 sample, a substantial
amount of cobalt is in ion-exchanged form, not reducible to metallic
cobalt by hydrogen at 350°C. This accounts for the negligible hy-
drogen uptake by the 2.7% Co-ZSM-5 sample. In samples with higher
cobalt loading, C0304 crystallites of average diameter -250 Aformed
162 v. U. S. RAO ET AL.
on the exterior of the zeolite [lS,16]. The co304 could be reduced
to hcp cobalt metal by flowing hydrogen at 3S0°C, and the resultant
specimens showed appreciable hydrogen uptake in the chemisorption
experiment. The magnetic studies showed a much higher degree of
reduction for the solution-impregnated samples with S.9% Co and 9.0%
Co than for the sample with 2.7% Co. The XPS studies on hydrogen-
reduced samples with about S% or more Co showed a considerable
amount of metallic cobalt, whereas samples with less than 3% Co
showed a high amount of Co 2+

The IR studies were conducted on Co-ZSM-S after pyridine had

been adsorbed on the samples. Such studies provide the relative
number of Br~nsted and Lewis acid sites from the examination of the
lS47 cm-l and l4S4 cm-l band intensities, respectively. Ion-ex-
changed cobalt increases the number of Lewis acid sites relative to
the number of Br~nsted acid sites [lS,16]. The physically admixed
metal-zeolite catalysts had the relative intensity of Br~nsted to
Lewis peaks unchanged from that in the parent H-ZSM-S. The increase
in the number of Lewis acid sites in the solution-impregnated samples
suggest that Co2+ ions have exchanged some of the protons in the
zeolite. The Co2+ ions act as additional Lewis sites and adsorb
pyridine coordinatively. Semiquantitative estimates show a decrease
in the number of Br~nsted acid sites per unit cell of ZSM-S from 4.S
to 1.1, and an increase in Lewis acid sites from 1.7 to 3.4, between
H-ZSM-S and solution-impregnated Co-ZSM-S. The decrease in the
number of Br~nsted acid sites can account for the lower aromatic
fraction in the synthesis gas conversion product from the solution-
impregnated catalysts in comparison to the product obtained from the
physically admixed catalysts (Table 1). It can also explain the
lower rate of aromatics formation in the conversion of ethylene
(Table 2) over ion-exchanged Co-ZSM-S in comparison to H-ZSM-S or
physically admixed Co-ZSM-S catalysts.

Ammonia adsorption studies [13] were performed on H-ZSM-S and

solution-impregnated Co-ZSM-S samples. The heat of ammonia adsorp-
tion on H-ZSM-S was in reasonably good agreement with the values
obtained from micro calorimetric studies [11]. The plots of the heat
of adsorption versus amount of ammonia adsorbed in the solution-
impregnated Co-ZSM-S samples exhibited anomalies that could be attri-
buted to the presence of ion-exchanged Co 2+ species. There was an
increase in the total adsorption capacity after Co2+ exchange, pos-
sibly from incipient complex formation of NH3 with Co2+.


The experiments show important differences in selectivity de-

pending on the method of preparing Co-ZSM-S catalysts for synthesis
gas conversion. Solution-impregnation methods result in catalysts
that yield a gasoline-range liquid hydrocarbon product containing a

smaller fraction of aromatics than that yielded by catalysts pre-

pared by the physical admixture method. The difference in selec-
tivity can be traced to the lowering of the Br~nsted acidity in the
solution-impregnated catalysts owing to the partial exchange of the
protonic acid sites by C02+ In the 5% to 9% Co concentration range,
the solution-impregnated catalysts provided a higher liquid hydro-
carbon yield than the physically admixed catalysts. These observa-
tions provide clues to the catalyst preparation method needed to
obtain a desired gasoline-type liquid product (for example, with
high or low percentage of aromatics).

Using metal-zeolite catalysts for synthesis gas conversion, it

appears possible to exceed the limitations to gasoline-range selec-
tivity (48%) imposed by Schulz-Flory kinetics. Some of the primary
F-T products such as C2-C4 olefins and oxygenates are converted to
gasoline-range components, including aromatics, by the catalytic
action and shape-selective function of ZSM-5. The heavier hydro-
carbons (Cll+) that are formed in the F-T reactions are cracked by
ZSM-5 to provide additional gasoline-range product.

The utility of ethylene conversion studies as a means of char-

acterizing the acid catalytic function is apparent. The studies can
also provide information on the rate of deactivation of the strong
acid sites responsible for the formation of aromatics. Investiga-
tions are required on the possible ways of modifying the zeolite to
retard the rate of deactivation. It is believed that the acid sites
are deactivated by coke or coke precursors originating from aromatic
compounds. It has been indicated that cracking reactions over BY,
H-ZSM-5, and H-ZSM-ll loaded with 1% Pt were nearly free from deac-
tivation [17]. A possible mechanism is the hydrogenation of the
hydrogen-deficient coke precursors by the Pt. It would be desirable
to examine the rate of deactivation in ethylene conversion and syn-
thesis gas conversion after the incorporation of about 1% Pt into
the catalysts examined in this investigation, and such studies are
being planned.

From the results and the tone of the discussion, it might appear
at first that metal-zeolite catalysts prepared by the physical ad-
mixture method may be superior to those prepared by solution-impreg-
nation methods, since the latter are likely to cause alterations in
the acid sites of the zeolite. The impregnation methods, however,
require further investigation, since they can provide ways to alter
the degree of metal dispersion in the catalyst. Metal carbonyls
[16,18] and metal-organic compounds are more likely to yield higher
metal dispersions on zeolite supports than solutions of metal salts.
Well-dispersed Fe-ZSM-5 [16,18] and carbon-supported iron [19] cata-
lysts with Fe particle size in the range 30 to 70 A show superior
maintenance of F-T activity with time. It has been reported [20]
that metal-exchanged (Fe 2+, NH4+)Y can be reacted with an anionic,
metal-containing coordination compound that is water-soluble, such
164 v. U. S. RAO ET AL.
as (NH4 )3[Fe(CN)6], to yield an insoluble compound, Fe3[Fe(CN)6]2,
distributed throughout the zeolite while the zeolite itself returns
to the ammonium form. The insoluble complex can later be reduced
in hydrogen to finely dispersed metal in the zeolite. It is thus
evident that impregnation and exchange techniques need to be inves-
tigated further to obtain metal-zeolite catalysts with highly dis-
persed metal components for the study of F-T and related reactions.


The authors would like to thank Bernard D. Blaustein and John

M. Stencel for helpful discussions. The technical support of John
T. Veres is deeply appreciated.


l. C. D. Chang, and A. J. Silvestri, J. Catal., 47:249 (1977).

2. P. Dejaive, J. C. Vedrine, V. Bolis, and E. G. Derouane,
J. Catal., 63:331 (1980).
3. W. K. Kaeding, and S. A. Butter, J. Catal., 61:155 (1980).
4. J. P. Van den Berg, J. P. Wolthuizer, and J. H. C. Van Hooff,
Proc. Fifth Int. Conv. on Zeolites, ed. by L. V. Rees, p 649,
Heyden, London, 1980.
5. C. D. Chang, W. H. Lang, and A. J. Silvestri, J. Catal., 56:268
(1979) •
6. P. D. Caesar, J. A. Brennan, W. E. Garwood, and J. Ciric,
J. Catal., 56:274 (1979).
7. V. U. S. Rao. and R. J. Gormley, Hydrocarbon Process., 59(11):
139 (1980).
8. T. J. Huang, and W. O. Haag, ACS Symp. Ser., 152:308 (1981).
9. R. J. Argauer, and G. R. Landolt, U.S. Patent 3,702,886 (1972),
Example 24; assigned to Mobil Oil Corp.
10. M. E. Dry in Catalysis - Science and Technology, ed. by J. R.
Anderson and M. Boudart, Vol. I, Ch. IV, P 159, Springer-
Verlag, Berlin, 1981.
11. A. Auroux, P. C. Gravelle, J. C. Vedrine, and M. Rekas, Proc.
Fifth Int. Conf. on Zeolites, ed. by L. V. Rees, p 443,
Heyden, London, 1980.
12. N. Y. Topsoe, K. Pedersen, and E. G. Derouane, J. Catal., 70:41
13. D. T. Hayhurst, private communication, and to be published.
14. K. Wicker, private communication, and to be published.
15. J. M. Stencel, V. U. S. Rao, A. G. Dhere, R. J. DeAngelis, and
K. H. Rhee, to be published.
16. V. U. S. Rao, Proc. Nordic Conf. on Surface Sci., Tampere, Fin-
land, August 1982 (to appear in Physica Scripta, 1983).
17. P. A. Jacobs, J. A. Martens, J. Weitkamp, and H. K. Beyer,
Faraday Discuss. Chem. Soc., 72:353 (1981).

18. R. T. Obermyer, L. N. Mu1ay, C. Lo, M. Oskooie-Tabrizi, and

V. U. S. Rao, J. App1. Phys., 53:2683 (1982).
19. H. J. Jung, P. L. Walker, and M. A. Vannice, J. Cata1., 75:416
(1982) •
20. J. Scherzer, and D. Fort, J. Cata1., 71:111 (1981).


F. G. Dwyer and W. E. Garwood

Mobil Research and Development Corporation

Paulsboro, NJ 08066


The intimate combination of zeolite ZSM-5 with CO reducing

metals for conversion of synthesis gas to gasoline boiling range
hydrocarbons has been previously reported. Chang et al [1] combined
the metal function with the zeolite in the same particle by impreg-
nation with metal nitrate solutions or mixing with hydroxide gel,
while Caesar et al [2] used a composite catalyst of separate 700 to
l500~m diameter particles of Z5M-5 and the Fischer-Tropsch component.
The latter paper showed that a mixture of potassium-promoted fused
iron catalyst with a volume excess of ZSM-5 generated a C5-200°C
aromatic gasoline with a research octane number of 92.

In this paper, the controlling effect of the Si02/A1203 ratio

of the ZSM-5 zeolite particles on the olefinicity of the hydrocarbon
product is described, along with its implications on overall mecha-
nism. The data are derived from a Mobil patent, [3] and similar
results have been reported by Rao et al [4] using metal-impregnated
ZSM-5 catalyst. In related work, Inui and Takegami [5] have physi-
cally blended ZSM-5 with methanol synthesis catalyst to form a com-
posite catalyst with activity for the conversion of synthesis gas to
liquid hydrocarbon. Also, Nijs et al [6] have reported on the chain
limitation of Fischer-Tropsch products in zeolites, using ruthenium
exchanged Y-molecular sieve.


The experiments were carried out in a high pressure microunit

having a 15/32" ID stainless steel reaction, described previous-


1y [7]. Gaseous product was analyzed hy mass spectrometry, and

liquid hydrocarbon product, after separation from the aqueous layer,
by gas chromatography using the open tubular GC columns and selec-
tive olefin adsorption procedures described by Bloch et a1 [8].

The ZSM-5 zeolite was synthesized by known procedures [9]. The

Fischer-Tropsch component was a commercial, potassium promoted, am-
monia synthesis catalyst obtained from Girdler under code name G-82.
Both components were sized to 14-25 mesh (700 to 150~m), mixed
before charging to the reactor, and pretreated in situ with hydrogen
overnight at 510°C. A premixed 2/1 H2/CO synthesis gas obtained from
Matheson was used as the charge at standard reaction conditions of
200 psig, 320°C bed inlet temperature (-330°C hot spot), and 3300
GHSV (1.5 \~SV) based on 4 volumes of zeolite plus one volume of
Fe-K (16,500 GHSV based on iron alone).


Effect of Ratio of Zeolite to F-T Component

In order to insure that the results of the experiments would

be reliable, it was necessary to establish an experimental procedure
that would give accurate and reproducible data. Previous work by
Caesar et a1 [2] had shown that a ratio of four volumes of ?eo1ite
to one volume of Fe-K is recommended to avoid wax formation and unit
plugging and hence give reproducible results. The results of their
work is summarized in Table 1.

It is striking that the 4/1 ZSM-5/Fe-K mix gives the same re-
sults when diluted fourfold with quartz, i.e. the ZSM-5 component
interacts with the Fe-K component even when separated from it by
many molecular distances. This will be discussed in more detail
later as it relates to mechanism.

Effect of Varying Si02/A1203 Ratio

The unique nature of the zeolite ZSM-5 makes it readily suit-

able for the study of the contribution of varying acidity levels to
the catalysis of this reaction. Unlike most other zeolites, ZSM-5
can be crystallized over a very wide range of Si02/A1203 ratios
extending to the essentially A1203-free species. No post crystalli-
zation extraction of A1203 is required, which could confound the
interpretation of the results.

Results of varying the ZSM-5 Si02/A1203 ratio from 70 to 500,

and then to 1600, are shown in Table 2.

Conversion of CO is 94% and higher in all cases, 68-69% going

to hydrocarbon product, and the remainder to C02 via the water-gas

Table 1. Effect of Ratio of HZSM-5 a to Fe-K

FeK, volumes 1 2.5 1 1

HZSM-5, volumes 0 2.5 4 4
Quartz, volumes 4 0 0 20

Time On-Stream, Days 1 2 2 4 2 5 2 5

CO Conv., wt % 76 97 95 94 98 90
H2 Conv., wt % 47 58 54 49 56 40
C5+ Product
Wt % of Total HC 40 52 W 53 61 56 58
Olefins, wt %
Aromatics, wt %
Trace P~
P~ 3
Oxygen, wt % 1.4 G 0.1

aSi0 2 / A1 203 70/1

shift reaction (CO + H20 + CO 2 + H2 ). The hydrocarbon composition

by carbon number is similar in all cases. All the liquid products
are in the gasoline boiling range, and have a high C5+ research
octane number of 92.

Olefin contents increase with increasing Si02/A1203 ratio, as

shown for C2 and C3 in Figure 1, C4 and C5 in Figure 2, and C6+ in
Figure 3. The C5's were -70% olefins, the highest of any carbon
number. In the C6+ product, aromatic content decreased with increas-
ing Si02/A1203 (Figure 4) as expected, due to decreased acidity. The
direct relationship of acidity to aluminum content has been previous-
ly reported by Meisel and Heisz [10].

Effect of Decreasing HZSM-5<Acidity with Potassium

In order to establish that the effect seen was definitely due

to the acidity of ZSM-5, a reduced acidity catalyst was prepared by
neutralization of the ZSM-5 acidity with potassium.

The 70/1 Si02/Al203 zeolite was exchanged with an aqueous solu-

tion of KN03 with results shown in Table 3.

The composition of the hydrocarbon product is very similar to

that from the 1600/1 Si0 2 /Al 20 3 zeolite in carbon number distribution
and C6+ aromatics, but somewhat lower in olefin content. The octane
number of the C5+ product is -92 in both cases, indicating that the
structure of the olefins must be very similar, i.e. highly branched.

Table 2. Varying Si0 2 /A1 20 3 Ratio of HZSM-5 Component with Volume

Ratio HZSM-5/Fe{K) = 4/1; Second Day on Stream.

Si02/A1203 Ratio 70 500 1600

CO Conversion, wt % 95 96 94
H2 Conversion, wt % 54 53 45

Wt % C Converted to
C02 32 31 31
Hydrocarbon , 68 69 69

Hydrocarbon Carbon No. Distribution, wt %

17 20 18
4 7 8
8 5 6
18 12 14
11 12 13
42 44 41

Liq. Product, 90% BP (OF) 402 358 356

C5+ O.N. (R + 0) 92.1 92.1 92.6

Comparison of Compositions with Equilibria

In an effort to further understand the role of acidity in the

chemistry of the reaction, we have looked at the composition of the
C5 olefins, which shows a high percentage of branched isomers (Table
4). Values are essentially the same regardless of Si02/A1203 ratio,
and are very close to equilibria. Similar results charging C2-C10
olefins directly over HZSM-5 have been reported [11].

Detailed composition of the C2+ product from 1600/1 Si02/A1203

HZSM-5 is listed in Table 5. Only trace amounts of hydrocarbon above
Cll were formed. Olefins contents peak in the C4-C6 range and de-
cline above C7 due to cyclization.

The C2-C6 data from Table 5 were used to calculate the mole %
distribution of the olefins for comparison with equilibria, shown
in Table 6 (equilibria data are available through C6 only) [12].
The partial pressure of the total olefins at the exit of the reac-
tor is about 6 psia, and the comparison is made at this pressure.

Ten times more ethylene is present in the product than should

be if equilibria were attained. In the work charging olefins direct-
ly to the reactor [11], it was found that ethylene production was
lower than that expected from equilibria calculations «1 versus 2%).






~ 200




10~ ___ 4_ _ _ __ L_ _ _ _ ~_ _ _ _~ _ _~ _ _ _ _ ~

o 10 20 30 40 50 60
Olefin Content, wt %

Figure 1. Effect of Si02/A1203 ratio on C2 and C3 olefin content

Mechanistic Implications

It is generally accepted that the mechanism of the Fischer-

Tropsch reaction is very complicated, such that there is no general
agreement, particularly on the initiation reactions [13]. The exper-
imental data discussed above are consistent with some aspects of the
chain propagation mechanism proposed by Henrici-01ive and Olive [14].
Regardless of how the initiating step and early propagation steps
occur, it is reasonable that eventually olefin species will be de-
sorbed and can be readsorbed on the metal as suggested by Henrici-
Olive and summarized below:

Desorption. R-CH3-CH2-M + R-CH2=CH2 + H-M

Readsorption and Further Growth.

R-CH=CH2 + H-M M

} + Further Growth.





q: 200
2 100


10~ __~____~__~~__~____~__~~__~.
o 10 20 30 40 50 60 70
Olefin Content, wt %

Figure 2. Effect of Si02/Al203 ratio on C4 and Cs olefin content

It has been proposed by Caesar et al [2] that these reactions

are accelerated at the higher temperature used in the mixed catalyst
system (330 vs 250°C for conventional F-T), but that instead of read-
sorbing on the metal, the olefins are intercepted by the zeolite.
The present work suggests that this occurs to a large extent with
formation of the first olefin formed, i.e. ethylene, and that the
metal site (iron) involvement in any further reaction is minimized.
The powerful attraction of the zeolite for the desorbed olefin is
evidenced by the results of diluting the catalyst mixture fourfold
with quartz, discussed earlier.


Physical mixtures of particles of high Si02/A1203 HZSM-S and

potassium-promoted iron have good activity for converting synthesis
gas to olefins. Results indicate that the zeolite intercepts de-
sorbed ethylene from the Fischer-Tropsch component, freeing its
surface to a large extent from further propagation reactions. Vary-
ing the acidity of the ZSM-5 component by changing the Si02/A1203
ratio or by addition of potassium controls the extent of the acid-
catalyzed olefin reactions.




o 10 20 30 40 50 60
c.+ 0lefln5, wt %
Figure 3. Effect of Si02/Al203 ratio
on C6+ olefin content.





cS 200



10~ __~____~__~~__~____~__~
o 10 20 30 40 50 60
Co' Aromatics, wt %

Figure 4. Effect of Si0 2 /Al 20 3 ratio on

C6+ aromatic content.

Table 3. Effect of Potassium on 70/1 Si0 2 /A1 203 HZSM-5 Acidity

70/1 Si02/A1203 1600/1 Si0 2 /A1 20 3

HZSM-5 No K 0.95 Wt % K No K

CO Conversion, wt % 95 97 94
H2 Conversion, wt % 54 59 45
H~drocarbon Carbon No. Distribution, wt %
C1 17 17 18
C2 4 6 8
C3 8 4 6
C4 18 11 14
C5 11 10 13
C+ 42 52 41
Olefin Content b~ C No. wt %
C2 6 26 40
C3 7 45 53
C4 4 53 68
C5 3 57 74

C6+ Aromatics, wt % 50 13 13
C6+ 01efins, wt % 1 22 37

C5+ ON (R + 0) 92.1 92.0 92.6

Table 4. C5 Olefin Composition

Si0 2 /A1 20 3 Ratio 70 500 1600 Equilibrium

(0.95% K)a 330°C

1-Cs= 3 3 2 2
2M1C4= 19 19 19 21
3M1C4= 2 2 2 3
trans-2Cs= 12 13 12 9
cis'-2Cs= 6 7 6 8
2M2C4= 58 56 59 57

aYie1d of C5 olefins from HZSM-5 (no K) too low for analysis.


Table 5. Composition of C2+ Hydrocarbon Product Produced over

1600/1 Si0 2 /Al 203 :HZSM-5
Composition, wt %
C No. Wt % 01efins Naphthenes Aromatics

2 10 40
3 7 53
4 17 18
5 16 74
6 15 62 3 1
7 12 59 8 1
8 11 32 -25 a 10
9 7 15 -50a 27
10 4 <1 89
11 1 <1

aLumped from GC analysis as C5 and C6 napthenes. Total unknowns

-1%, remainder paraffins throughout C No. range.

Table 6. Comparison of C2-C6 01efins with

Equilibria, Mole %; 1600/1 Si021

Found 6 psia

C2= 20 2
C3_= 12 12
C4- 29 35
c5= 24 36
C6= 15 15
100 100


The microunit operation by B. T. Bickness, GC analysis by

J. H. Stockinger, and helpful discussions with C. R. Kennedy are
acknowledged with thanks.


1. C. D. Chang, W. H. Lang, and A. J. Silvestri, J. Cata1., 56:268

2. P. D. Caesar, J. A. Brennan, W. E. Garwood, and J. Ciric, ~
Cata1., 56:274 (1979).
3. F. G. Dwyer and W. E. Garwood, U.S. Patent 4,172,843 (1979);
assigned to Nobi1 Research and Development Corp.
4. V. U. S. Rao, R. J. Gormley, H. W. Penn1ine, L. C. Schneider,
and R. Obermyer, Preprints, Div. Fuel Chern., ACS, 25:119
(1980); V. U. S. Rao and R. J. Gormley, Hydrocarbon Process.,
59(11):139 (1980); and V. U. S. Rao and R. J.Gorm1ey, U.,S.
Patent 4,340,503 (1982); assigned to U.S. Department of
5. T. Inui and Y. Takegami, Preprints, Div. Pet. Chern., ACS,
27(4):982 (1982).
6. N. H. Nijs, P. A. Jacobs, and J. B. Uytterhoeven, J. Chern.
?oc., Chern. Commun., 180 (1979).
7. N. Y. Chen and W. E. Garwood, Adv. Chern. Ser., 121:575 (1973).
8. N. G. Block, R. B. Callen, and J. H. Stockinger, J. Chromatogr.
Sci., 15:504 (1977).
9. R. J. Argauer and G. R. Landolt, U.S. Patent 3,702,866 (1972)
and F. G. Dwyer and E. J. Jenkins, U.S. Patent 3,941.871
(1976)[RE 29.948(1979)]; assigned to Nobi1 Research and
Development Corp.
10. S. L. Neise1 and P. B. Weisz. Adv. Cata1. ~hem. II Symp.,
University of Utah. Salt Lake City. Nay 1982.
11. lv. E. Garwood. Preprints, Div. Pet. Chern., ACS. 27(2) :563
12. E. D. Rossini. "The Science of Petroleum." Vol. V. Part I.
Oxford University Press (1950).
13. R. B. Anderson. in "Catalysis." Vol IV. ed. by P. H. Emmett.
Reinhold. New York. Chap 1-3 (1956); T. T. Tsotsis. V. U. S.
Rao. and L. N. Polinski. AIChE J •• 28(5):847 (1982).
14. G. Henrici-01ive and S. Olive. Angew. Chern •• Int. Ed., 15(3):
136 (1976). .


J. G. Goodwin, Jr.,* Y. W. Chen,a and S. C. Chuang

Chemical and Petroleum Engineering Department

University of Pittsburgh
Pittsburgh, PA 15261


While the catalytic activity .of a catalyst is important, of even

more importance is catalytic selectivity. This is especially true
for the Fischer-Tropsch (F-T) synthesis, where much work is being
done to selectively make gasoline, olefins, or other classes of hy-

One way to affect the selectivity of a particular F-T metal cat-

alyst is by its support. Much research has been reported for many
metals supported on Si02, A1203, various zeolites, and the newer SMSI
class of supports. However, little work has been reported for the
F-T synthesis over catalysts supported on many of the other metal
oxides. While many of these oxides have the disadvantage of low sur-
face area, they may be able to produce interesting modifications in
catalytic behavior when used as supports.

Ruthenium is well known to be one of the most active catalysts

for the F-T synthesis. It is able to produce higher hydror.arbons up
to C7 at even one atm pressure. In addition, Ru is easily reduced
and does not form carbides under F-T conditions. These facts make
Ru an ideal F-T metal to use to probe support-metal effects.

Numerous studies have been reported concerning the use of A1203,

Si02, Ti02, and zeolite-supported Ru for the F-T synthesis. Little
has been published, however, concerning F-T synthesis over Ru sup-

aCurrent address: SRI International, Menlo Park, CA

180 J. G. GOODWIN, Jr. ET AL.

ported on other oxide supports such as ZnO, Cr203, etc. Kimura et

ale [1] studied Ru supported on Si02, ZnO, y-A1203, and MgO using IR
spectroscopy. Tamoyuki and Mesaki [2] have reported the results of
a study of methanation from CO or C02 over a supported Ni-La203-Ru
catalyst. No reaction data has been found in the literature for F-T
synthesis over Ru supported on ZnO or La203.

This paper reports on a study of metal-support effects in Ru

supported on Ce02, Cr203, La203, MgO, Si02, SnO, Th02, ZnO, and Zr02'


The catalysts were prepared by the incipient-wetness technique

using RuC13·1.5H20 (obtained from Strem Chemicals, Inc.) dissolved
in distilled water to form a solution having a concentration suffi-
cient to yield the proper metal loading when 0.7 ml of solution was
used to wet each gram of support. A series of 9 supports were used.
The sources of these supports and their approximate surface areas are
shown in Table 1. After the impregnation, the samples were dried
overnight in air at 40°C and then bottled. Prior to reaction, the
catalysts were reduced in flowing H2 by heating to 400°C in stages
over 23 hours and holding at that maximum temperature for 2 hours.

The determination of average ruthenium particles sizes for a

number of catalysts was made using x-ray diffraction (XRD) with a
Mo Ka radiation source.

Fischer-Tropsch synthesis was carried out using a differential

reactor system. The gases used for these reaction studies were ob-
tained from Air Products and Chemicals, Inc. and included H2 (UHP,
99.999%), He (UHP, 99.997%), and a H2/CO mixture (H2/CO = I, UHP,
99.9%), which were purified by passing through molecular sieve and
drierite traps to remove water and metal carbonyl contaminants.
Prior to passage through the molecular sieve trap, the hydrogen was
passed through a Deoxo unit to react the oxygen to form water. The
reaction was carried out at 101.3 kPa (1.0 atm) total pressure and at
temperatures in the range 220-320°C. By alternating short reaction
periods and longer reduction periods, called the "hydrogen bracketing
technique", stable catalyst activity could be achieved after several
cycles. The "hydrogen bracketing technique" did an exceptionally
good job of maintaining a clean Ru surface and giving reproducible
results. High space velocities of 1000 to 3000 hr -1 were typically
used to keep CO conversions around 5% or less. This provided con-
version data from the differential reactor, minimized heat and mass
transfer effects, eliminated any significant effects due to product
inhibition, and minimized complications from secondary reactions. An
onstream Perkin-Elmer Sigma IB gas chromatograph was used to analyze
the reactor effluent. Products were separated using a 6 ft x 1/8
inch stainless steel column packed with Porapak Q and were detected

Table 1. Properties of the Supports [3]

Approximate Redox
Metal -LlHf
Source Surface Area Potential
Oxide (mZ/g) (kcal/mol)

CeOZ Strem 99 Z33.0 1.443

CrZ03 AHa ZOO Z69.7 -0.41
Laz03 AHa 14 458 -Z.37
MgO Strem 50 143.8 -Z.375
SiOZ Strem 350 Z05.0 -0.84
ThOZ AHa 10 Z9Z -1.9
ZnO Mallinckrodt ZO 83.Z -0.763
ZnOz Strem 50 Z58.Z -1.43

by both thermal conductivity (TCD) and flame ionization (FID) de-



Reaction results from this study are presented in Tables Z and

3 and Figures 1 and Z. It is obvious from the data shown that the
species of support had a strong influence on the catalytic activity
and selectivity. The specific activity (per g of catalyst) decreased
in the order: Ru/CrZ03 » Ru/MgO > Ru/ZrOZ > Ru/CeOz > Ru/ThOZ >
Ru/SiOZ > Ru/LaZ03 » Ru/ZnO > Ru/SnO. However, based on the XRD
determination of average Ru particle size, it would appear that Ru/-
LaZ03 was one of the most active catalysts on a per site basis. The
Ru/ZnO catalyst was one of the least active on this basis. The site
activity of Ru/ZnO is estimated to be two orders of magnitude lower
than that of Ru/Laz03. In early work on the promotion of Co-kiesel-
guhr catalysts, Crz03 was found to lower the F-T activity while MgO
and ZnO had little influence [4]. In a study of supported FeCu at
5ASOL [5], CrZ03 and MgO supports tended to lower activity about 10%
to 50%, respectively, relative to the unsupported catalyst, while
ZnO increased the activity comparable to that of the SiOZ-supported
catalyst. In this study of Ru, however, ZnO appears to have greatly
reduced the activity on the per site basis. The other supports with
the possible exception of SnO, do not appear to have affected activ-
ity nearly as much.

The methane selectivity decreased in the following order:

Ru/SnO > Ru/MgO > Ru/ThOZ > Ru/Crz03 > Ru/CeOZ > Ru/SiOZ > Ru/ZrOz >
Ru/ZnO > Ru/Laz03. Figures 1 and Z show the effect of temperature
182 J. G. GOODWIN, Jr. ET AL.

Table 2. Kinetic Behavior of Supported Ru Catalysts

dpa rCOb ECH4

Catalyst EyO
(nm) (llmole/g.sec) (kcal mole) (kcal/mole)

3% Ru/ Cr 20 3 < 2.0 4.2 22 27

6% Ru/ Cr203 < 2.0 3.2 20 24
3% Ru/MgO < 2.0 1.12 18 25
3% Ru/Zr02 1.04 16 20
3% Ru/Th02 < 2.0 0.88 17 20
3% Ru/Si02 0.81 17 22
3% Ru/Ce02 0.95 12 17
3% Ru/La203 13.0 0.68 16 18
3% Ru/ZnO < 2.0 0.07 26 28
3% Ru/SnO 0.01 25 25

aAverage Ru particle diameter determined by X-ray diffraction.

bReaction conditions: H2/CO = 1, 250°C, 1 atm, GHSV = 1200.

of reaction on selectivity. It is interesting to note that the two

catalysts producing abnormally high fractions of olefins (Ru/ZnO and
Ru/La203) also produced longer-chain hydrocarbons. The methane se-
lectivity increased with temperature in all cases; however, it was a
weak function of temperature for ZnO and La203 supported Ru. Even
at 320°C, methane constituted only 26 wt % of the hydrocarbons formed
on Ru/La203. The olefin selectivities for these two catalysts were
essentially temperature-independent and were well above 80% in the
C2-C4 fraction. It should be pointed out that catalysts with lower
activities did not necessarily show a shift in product distribution
towards methane, and catalysts with high methane selectivities were
not necessarily high activity catalystti.

The apparent activation energy for this reaction varied widely

from support to support as shown in Table 2. In each case, the acti-
vation energy of methanation was larger than that of hydrocarbon syn-
thesis. This confirms that methane is a favorable product at higher
temperatures. There is no obvious correlation between specific ac-
tivity and activation energy.


As can be seen from the results, the catalytic properties of Ru

for F-T synthesis vary greatly depending on the support used (activ-
ity: more than 2 orders of magnitude different, CH4 selectivity:
25-85 wt %, and olefin selectivity: 0-90% of C2-C4). It is of great

Table 3. Product Distribution of Supported Ruthenium Catalysts a

Product Distribution (wt %)

Cl C2= C2 C3= C3 C4= C4 C5+

3% Ru/Cr203 45.5 1.3 10.3 16.1 6.8 4.7 7.9 7.4

6% Ru/Cr202 44 1.2 11.6 17.9 4.6 4.0 7.2 8.7
3% Ru/MgO 55.8 0.5 12.5 5.3 7.7 0.2 9.7 7.6
3% Ru/Zr02 39.8 28.6 10.6 17 .8 6.7 3 9.5 9.2
3% Ru/ Th02 44.8 2.4 10.9 17.3 5.7 5.1 5.4 7.4
3% Ru/ Si02 38.8 1.8 11.1 20.7 5.6 4.7 9.7 7.6
3% Ru/ Ce02 39.1 4.2 7.4 18.8 4.1 9.6 5 11.2
3% Ru/La203 21.4 18.3 2.2 24.5 0.8 14.5 0.8 16.6
3% Ru/ZnO 27.7 12.4 2.4 28.8 2.7 15.4 2.4 8.2
3% Ru/SnO 98 1 1

1, 250°C, 1 atm, GHSV 1200.

interest to understand how such great differences in properties are


A number of properties of oxide supports can be proposed that

might be related to the generation of metal-support interactions.
The more obvious ones are acidity/basicity, redox potential, reduc-
ibility, semiconductor (type)/insulator, d-orbital characteristics
of the support, and the presence of impurities.

It is known that the acidity (or basicity) of the support may

affect the catalytic properties of a supported metal. In this case,
there appears to be no obvious role of acidity/basicity in the deter-
mination of these properties. The two strong bases, MgO and La203,
produced the extremes in selectivity. The other, amphoteric supports
exhibited properties in between those of MgO and La203'

In an extensive study of methanation over supported Ni cata-

lysts, Trimm and Karal [6] reported that there is a very good corre-
lation between methanation activity and redox potential of the sup-
port. However, the results for Ru do not correlate with the redox
potential of the supports. The methane selectivity and specific ac-
tivity also do not correlate well with the heat of formation of the
oxide per oxygen atom.

Tauster et ale [7] have suggested that SMSI properties are re-
lated to the reducibility of the support. They found this criterion
to hold for 12 out of 13 transition metal oxides investigated; only
184 J. G. GOODWIN, Jr. ET AL.


w 70
~ 60
#- 50

5 40

3: 30


200 220 240 260 280 300 320 340 360
Figure 1. Effect of the support on methane selectivity.

Cr203 did not fit the correlation. Table 4 gives the reducibilities
of the supports investigated in this study, as well as the metal
oxides (MnO, Nb20S' and Ti02) found by Tauster et al. [7] to have the
strongest metal support interactions. One might expect ZnO to be an
excellent candidate for SMSI. The reaction results in this study


5 80


y 60
5 40
~ 20
W \
o S!. -... MgO
o 'Bl - .....
200 220 240 260 280 300 320 340 360
Figure 2. Effect of the support on olefin selectivity.
Table 4. Reducibilities of the Supports "'tI
Metal Oxide Reduction to Reducibility [8] at 1000 0 K -t
SMSI Activity [Ref] ::z::
log (PH20/PH2) m
Th02 Th -18 m
Ce02 Ce -17 .6 .!.oj
MgO Mg -14.5 no [7] ::0
La203 La -13.7 CJ)
Zr02 Zr -13.6 no [7] n
Si ::z::
S102 - 9.1 no [7] CJ)
MnO Mn - 6.2 very strong [7] -<
Cr203 Cr - 5 little [7] -t
ZnO Zn ::z::
- 2.9 m
Nb205 Nb02 - 1.1 very strong [7] c:;;
Ti02 Ti305 - 1 strong [7]
Sn02 Sn + 0.2

186 J. G. GOODWIN, Jr. ET AL.

tend to confirm the possibility of SMSI with Ru/ZnO since the activ-
ity and CH4 selectivity were lower and the olefin selectivity greater
for Ru/ZnO relative to Ru/SiOZ. Such behavior is similar to that
found for SMSI Ru/TiOZ [9]. Our results also confirm those of
Tauster et al. [7] that, although CrZ03 would be a good candidate
for SMSI, it does not seem to occur. Ru/CrZ03 gave results in this
study very similar to those for Ru/SiOZ. Finally, although SnO is
the most reducible of all the oxides listed, it produced primarily
CH4. This may be explained by the fact that, if an oxide is reduced
too much, alloy formation may occur between the reduced metal of the
support and the catalytic metal. In the case of Ru/SnO, since Sn is
not active for F-T, formation of RuSn particles would be expected to
decrease the activity and increase CH4 formation. This is indeed
what our results show.

Recently, Chen and White [10] have suggested that SMSI behavior
may occur with those supports that have relatively high electrical
conductivities and work functions that are lower than the supported
metal. Comparing the conductivities of the inorganic oxide supports
in this study [11], it can be seen that CrZ03 and ZnO have the high-
est conductivities. However, as pointed out before, ZnO and Crz03
show significant differences in activities and selectivities of meth-
ane and olefin formation. A possible explanation for this difference
may lie in the fact that ZnO is an n type semiconductor and Crz03 is
a p type semiconductor [lZ]. However, it should be noted that Laz03
and Crz03 show significantly different properties although both are
p-type semiconductors. In addition, while Cr 203 has a higher con-
ductivity than SiOZ (SiO Z is an insulator), toe catalytic selectiv-
ities for Ru/cr Z0 3 and Rfi/Si0 2 are very similar.

Based on reducibility and electrical conductivity one would not

expect La203 to exhibit SMSI. Certainly, the F-T site activity of
Ru/LaZ03 is equal to or greater than that of Ru/SiOZ. Its selectiv-
ities are more comparable, however, with those of Ru/ZnO. In fact,
its CH4 yield is considerably below that of Ru/ZnO at high tempera-
tures. This seemingly anomalous behavior of La203 relative to the
other metal oxide supports may be due to the fact that it is more
prone to form a stable superficial carbonate [7]. This carbonate
might have some type of promotion effect on the Ru.

The comparison based on conductivity and type of semiconductor

does not give any correlation as illustrated. Although the individ-
ual cases such as ZnO and TiOZ are interesting, no general operation-
al theory has emerged. In fact, results of these studies are con-
tradictory and confusing. It is really too much to expect bulk con-
ductivity characteristics to correlate precisely with catalytically
important surface characteristics. Furthermore, the complex inter-
action of foreign atoms and pretreatment procedures for catalysts
makes the problem even more difficult.

It is beginning to become apparent that SMSI may just reflect

catalyst behavior rather than be due to a unique cause. SMSI has
now been reported for catalysts having the conventional supports of
Si02 [13-15] and Al203 [16-19]. In addition, Haller et al. [20] have
recently reported the existence of SMSI in Rh/MgO catalysts when less
than high purity (99.999%) MgO was used. Thus, the presence of im-
purities may produce SMSI in catalysts containing a support that is
essentially not reducible.

Since many semiconductors are oxides of transition-metals of

the fourth period, it might be reasonable to expect some correlation
with the character of the d-shells. The approach to this issue has
been through the use of crystal field theory and ligand field theory
[21-23]. It is postulated that during adsorption (or contact) on a
cation of the support, a change in coordination can occur. In most
such transformations the stabilization energy is negative, which
tends to increase the stability of the complex. For oxide supports
with cation configurations of dO - d5 and dID, the stabilization
energy is zero. Thus, oxides with cations of this type would be ex-
pected to have high activation energies. Maxima in stabilization
energy differences are between dO - d 5 , and d 5 - dID. Based on such
concepts, the results on Ti02 (dO), MnO (d 5 ), and ZnO (dID) seem to
indicate that the oxide supports with cation configurations of dO,
d 5 , and dID could exhibit the SMSI effect. However, V203 is also an
SMSI support but has a cation configuration of d 2 • In the fifth
period, neither Zr02 nor Y203 has SMSI properties but both have a
cation configuration of ciO. Thus, there appears to be no obvious
correlation of SMSI behavior with electronic properties.

Finally, while correlations of catalytic properties to a partic-

ular property of the support can be sought, it is very likely that
combinations of properties may be more of a factor in determining the
catalytic behavior than anyone single property. Although contro-
versy exists as to the precise cause of metal-support interactions in
general and SMSI in particular, the fact that the support can bring
about tremendous modifications in the catalytic properties of a sup-
ported metal is obvious.


The results of this study illustrate how greatly the catalytic

properties of a metal can be modified by a support. Depending on the
support used, the activity of Ru for F-T varied by as mucp as 2
orders of magnitude. Methane selectivity went from 21 to 98 wt %
and olefin selectivity ranged between ca. 0 and 90% of the C2-C4 hy-

The most interesting supports studied were ZnO and La203 since
188 J. G. GOODWIN, Jr. ET AL.

they were able to produce low CH4 yields and very high (> 80%) olefin
yields. Ru/La203 was one of the most active catalysts (on a site
basis), whereas Ru/ZnO was one of the least active. The results sug-
gest that ZnO may be an SMSI support.

While ZnO and La203 are not high surface area supports as is
normally desirable, the unique catalytic properties for F-T synthesis
of Ru in interaction with thern suggest interesting possibilities for
new catalyst designs.


Funding for this research was provided by the U.S. Department

of Energy, Office of Fossil Energy, under grant no. DE-FG22-82PC-


1. T. Kimura, J. Okuhara, M. Misono, and Y. Yoneda, Nippon Kagaku

Kaishi, 162 (1982).
2. I. Tamoyuki and F. Masaki, Chern. Lett. (Japan), 251 (1978).
3. D. L. Trimm, "Design of Industrial Catalysts," Elsevier, Amster-
dam, pp 267-287 (1980).
4. F. Fischer, and H. Koch, Brennst.-Chem., 13:61 (1932).
5. M. E. Dry, in "Catalysis: Science and Technology," Vol. 1, ed.
by J. R. Anderson and M. Boudart, Springer-Verlag, New York,
pp 159-255 (1981).
6. D. L. Trimm and E. Kara1, Studies in Surf. Sci. and Cata1., 7:
517 (1981).
7. S. J. Tauster, S. C. Fung, R. T. K. Baker, and J. A. Horsley,
Science, 211:1121 (1981).
8. T. B. Reed, "Free Energy of Formation of Binary Compounds," MIT
Press, Cambridge, Mass., pp 8-9 (1971).
9. C. H. Yang, Ph.D. Dissertation, University of Pittsburgh, 1983.
10. B. H. Chen and J. M. White, J. Phys. Chern., 86:3534 (1982).
11. G. V. Samsonov, "The Oxide Handbook," Plenum Press, New York,
12. S. R. Morrison, CHEMTECH, 570 (Sept. 1977).
13. G. R. Wilson and W. K. Hall, J. Cata1., 24:306 (1972).
14. R. L. Moss, D. Pope, B. J. Davis, and D. H. Edwards, J. Cata1.,
58:206 (1979).
15. H. Pra1iaud and G. A. Martin, J. Cata1., 72:394 (1981).
16. L. Gonza1ez-Tejuka, K. Aika, S. Namba, and J. Turkevich,
J. Phys. Chern., 81:1399 (1977).
17. F. M. Dautzenberg and H. B. M. Wolters, J. Cata1., 51:26 (1978).
18. K. Kunimori, T. Okouchi, and T. Uchijima, Chern. Lett. (Japan),
1513 (1980).
19. K. Kunimori, Y. Ikeda, M. Soma, and T. Uchijima, J. Cata1.,
79:185 (1983).

20. G. L. Haller, D. E. Resasco, and J. Wang, Preprints Div. of

Pet. Chern. (ACS), 28:559 (1983).
21. D. A. Dowden and D. Wells, "Proc. 2nd Intern. Congr. Catal.,"
Editions Technip, Paris, p 1499 (1961).
22. J. Haber and F. S. Stone, Trans. Farad Soc., 59:192 (1963).
23. S. R. Morrison, "The Chemical Physics of Surfaces," Plenum
Press, New York, 1977.


F. J. Waller

Central Research & Development Department

E. I. du Pont de Nemours and Company
Experimental Station
Wilmington, Delaware 19898


The incorporation of CO via metal carbonyls into various unsat-

urated substrates was first developed by Reppe during the thirties
and forties [1]. Rohm and Haas, BASF, and Dow Badische have success-
fully utilized modified Reppe technology in the manufacture of acryl-
ic acid (Equation 1) from acetylene [2]. Using Ni propionate as a

HC = CH + CO + ROH Ni(CO)4. H2C=CHC02R (1)

catalyst precursor, BASF also produces propionic acid (Equation 2)

H2C=CH2 + CO + ROH + CH3CH2C02R (2)

from ethene and CO [1,2]. However, with alpha-olefins, two isomeric

carboxylic acid derivatives are possible depending upon the transi-
tion metal catalyst. When R = CSHll' the alpha-olefin carboxylation
route (Equation 3) to linear synthetic fatty acid esters is cata-

~ RCH2CH2C02Rl
RCH=CH 2 + CO + R10H (3)

lyzed by phosphine-modified palladium(II)-group 4B metal halide com-

plexes [3]. The regioselectivity to the branched isomer, R = CH3'
is controlled by both aryl arsine and phosphine-modified palladi-

194 F. J. WALLER

um(II)-group 4B metal halide complexes [3]. The regiose1ectivity

to the branched isomer, R = CH3 , is controlled by both aryl arsine-
and phosphine-modified palladium(II) complexes [4,5]. For example,
propene yields predominantly (>90 mol %) methyl isobutyrate, a com-
pound of interest as a possible precursor to methyl methacrylate.

Palladium complexes have found extensive application of carbo-

alkoxylation and hydrocarboxylation catalysts [6-10]. Catalytic
carboxylation of olefins with turnover numbers >50 (mol product/mol
catalyst) and operating under mild temperature and pressure condi-
tions usually require an acid, HX, as a cocata1yst with a palladium
catalyst. These catalysts are either phosphine palladium(II) or
Pd(O) complexes [6,7]. Two mechanisms have been proposed for the
carboa1koxylation of olefins. The hydride mechanism is shown in
Scheme 1.

The trans-Pd(COPr-n)C1(~3P)2 complex, where ~3P represents tri-

phenylphosphine. that is an active catalytic specie in the hydroes-
terification tends to support the hydride mechanism [11]. Catalyst
activation via L2Pd is not unreasonable since these types of phos-
phine-modified complexes, in particular bis[tri(ortho-tolyl)phos-
phine], have been reported in the literature [12.13]. Kinetic stud-
ies on the carbonylation of cyclohexene (Equation 4) have been em-


ployed to partially support the above mechanism [14]. The authors

proposed a mechanism involving a slow step in the overall sequence
of events. Derived rate and equilibrium constants support the pro-
posed reaction pathway. However, an alternative mechanism has been
reviewed and is illustrated in Scheme 2 [15]. The complex. (03P)2-
PdC1(C02CH3), has been well characterized, and it is not unreason-
able to expect o.lefin insertion into the Pd-C bond [16].

The role of an acid, as a co.cata1yst. in the carbo.metho.xylation

reactio.n is different in each mechanism. It either functio.ns in
the fo.rmation of a palladium hydride Dr cleavage o.f the Pd-C bond,
respectively. catalysts are also. known for a few of the analo-

gous so.luble palladium co.mplexes [17J. Palladium chloride has been
bound to styrene resins via ~2P groups and used as a catalyst for
the ethoxycarbonylation of l-pentene. The resin-bound palladium
chloride catalysts yielded linear/branch ratios o.f -17.

The objective of this work was two-fold. First, we wished to

investigate the possible contro.1 of catalytic activity by changing
from a coordinating to non-coo.rdinating anion through the use of a
series of non-hydrogen halide acids and second, to compare homoge-
neous vs. heterogeneous catalysis.

Scheme 1. Proposed hydride catalytic cycle for carboalkoxylation.

Square brackets have been placed around characterized



Triflic, CF3S03H, and fluorosulfonic acid were distilled prior

to use in the carbonylation reactions. The perfluorinated acid,
C8F17S03H was sublimed from a solution of C8F17S03K and anhydrous
H2S04. Nafiona , a perfluorinated-ion-exchange polymer (PFIEP), was
in the powder form and had an equivalent weight of 1100. The acid
form of the resin was exchanged with an aqueous palladium nitrate
solution at 60°C for five hr. The filtered resin was dried under
vacuum at 110°C for 8 hr to yield a reddish-black powder.

Product Identification and Analysis

All GLC analyses were performed on a Hewlett Packard S700-A gas

chromatograph. The isomeric esters were separated on a 10' x 1/8"
GC column packed with 10% SE-30 on 80/100 mesh Supelcoport and

aDu Pont registered trademark. This material is no longer available

in powder or cube form.

Scheme 2. Proposed labile alkoxy catalytic cycle for carboalkoxyla-

tion. Square brackets have been placed around character-
ized complexes.

temperature programmed from 110°C to 220°C. The He flow was 30

ml/min. Molar response factors were referenced against toluene, an
internal standard. Reaction products were generally identified by
comparison with GC retention times of authentic samples and a GC mass

Carbomethoxylation Procedure

In the general carbomethoxylation procedure, a charged 330-ml

stainless-steel shaker tube was connected to a pressure cell. A
pressure vs. time curve was recorded during the reaction time of 1
hr at an initial total pressure of 13.8 MPa (2000 psi). The curve
was used to calculate the initial rate of carbomethoxylation at low-
olefin conversions. The rate was normalized for the moles of Pd
initially charged to the reactor. Units for the initial turnover
rate are mol ester/mol Pd/hr.

Several carboxylation experiments were conducted at constant CO

pressure by repressuring the reaction vessel to a total pressure of
13.8 MPa. When the olefin was cis-2-butene, two identical experi-
ments yielded initial turnover rates of 541 and 552, respectively.
These results suggest that the precision is reasonable with the type
of vessel employed.


The non-constant CO pressure experiments were carried out with

(03P)4Pd, 5% Pd/C, and the soluble acids H2S04' pMeC6H4S03H, CH3S03H,
FS03H, CSF17S03H, and CF3S03H. The heterogeneous acids consisted
of Dowex a , Amberlyst a 15, PFIEP, and a palladium-cation partially
exchanged PFIEP. The alcohol, CH30H, was used as the reactant and
solvent. Triphenylphosphine was added to prevent catalyst decompo-
sition to metallic palladium. The molar ratios of reactants are
specified at the bottoms of the tables. Conversions to the esters
were calculated as a ratio of mols ester/mols olefin compound charged
to the reactor. The selectivity to the linear ester is 100%, 71-74%,
and 73% for ethene, propene, and l-hexene, respectively. All reac-
tions were carried out at 100°C.

Under the mild temperature and pressure conditions of 100°C and

2000 psi, respectively, absence of the acid cocatalyst did not pro-
duce the carboxylic acid esters with either ethene or propene as
the unsaturated olefin and (03P)4Pd as the catalyst precursor.
These results are to be contrasted with reported catalytic activity
for the palladium complex (03P)2PdC12, either in the presence or ab-
sence of HCl [6,7].

Propene is carbomethoxylated to two isomeric esters, methyl

butyrate and methyl isobutyrate. Table 1 shows the effect on the
initial turnover rate using a series of soluble acids. The catalyst
precursor was 5% Pd on carbon or (03P)4Pd. With the catalytic system
described here, soluble perfluorinated sulfonic acids increase the
activity of palladium. There is a rate enhancement of 2.2 between
H2S04 and a triflic acid with {03P)4Pd. The linear isomer selectiv-
ity did not change over a range of acid strength. Therefore, the
coordination ability of the anion apparently does not have a pro-
nounced selectivity effect on the active catalytic intermediate. The
other salient feature is the decreased amount of sulfonic acid re-
quired to maintain high catalytic activity when compared with anhy-
drous HCl.

The carbomethoxylation of propene with (03P)4Pd under conditions

outlined in Table I, except at constant CO pressure, gave an activity
ratio of 1.7 based upon ester product analysis for the cocatalysts
CF3S03H and H2S04. This is .in good agreement with the comparison
of the rates by each acid above.

Similar results are obtained for the heterogeneous acids repre-

sented in Table 2. PFIEP is a perfluorinated ion-exchange polymer
with the general chemical structure shown below [IS]. The backbone

aDow, Rohm and Haas registered trademark respectively.

198 F. J. WALLER

Table 1. Effect of Soluble Acids on Carbomethoxy1ation of Propene

Catalyst Turnover
Acid (mequiv) Conv.
precursor a rate

5% Pd/C anhyd. HC1 (55.6) 1.1 12

5% Pd/C H2S04 (2.28) 26.2 349
5% Pd/C CF3S03H (2.29) 54.8 1498
(~3P)4Pd pMeC6H4S03H (2.30) 27.3 286
(~3P)4Pd H2 S04 (2.24) 22.2 382
(~3P)4Pd CH3S03H (2.29) 25.9 550
(~3P)4Pd FS03H (2.39) 31.6 656
(~3P)4Pd C8 F17 S03 H (2.26) 37.5 678
(~3P)4Pd CF3S03H (2.27) 47.4 828

a5% Pd/C (0.434 mmo1 Pd), ~3P/Pd = 15.4, 500 mmo1 MeOH, 333 mmo1
propene: (~3P)4Pd (0.433 mmo1 Pd), ~3P/(~3P)4Pd = 11.5, 500 mmo1
MeOH, 333 mmo1 propene.

of these polymers is quite chemically resistant and are thermally

stable up to 180-200°C or higher depending on the counter ion form.
PFIEP shows a rate enhancement of 5.2 over Amber1yst 15.


An interesting feature of the catalytic system with CF3S03H is

the decrease in measured turnover rate obtained for various olefins
studied. Table 3 summarizes the results with four olefins. If the
kinetic model [14] proposed by other researchers is correct, there
is a slow step in the overall transformation shown in Equation 4.
However, the net reaction in Equation 4 consists of three fundamental
steps: ligand dissociation, olefin coordination, and hydride addi-
tion across the olefin. It is possible that the non-coordinating
triflate anion opens up a vacant coordination around the central
metal for the olefin. The olefin structure has a pronounced influ-
ence on its binding to the metal. This suggests that olefin coor-
dination is the rate-determining step in the carbomethoxy1ation
reaction when the anions are less nucleophilic than other nearby

Table 2. Effect of Heterogeneous Acid on Carbomethoxylation of


Catalyst Turnover
Acid b (mequiv) Conv.
precursor a rate

Amberlyst 15 (2.49) 4.6 139

Dowex (2.31) 2.5 146
PFIEP (2.26) 23.1 719

a(03P)4Pd (0.433 mmol Pd), 03P/(03P)4Pd = 11.5, 500 mmol MeOH, 333
mmol propene.
bAmberlyst 15 is a macroreticular sulfonated polystyrene ion-exchange
resin, and Dowex is a sulfonated polystyrene ion-exchange resin.

ligands of CO and 03P. However, the limited examples cannot distin-

guish if the substituent effects on alkyl ethylenes are steric or
electronic in nature.

Slightly different turnover rates are obtained with the cocata-

lyst, PFIEP, and the olefins ethene, propene, and l-hexene. The
data in Table 4 reflect a decrease in turnover rate. The slowest
reacting olefin is l-hexene. In PFIEP, there are regions consisting
of a hydrophobic fluorocarbon phase with hydrophilic ionic areas
[18]. Apparently after the catalytic active specie, HPd(03P)n03SCF2-
~, is formed in the fluorocarbon matrix, there is a diffusional
barrier for large olefins. Separate but identical experiments in-
volving the protonation of amines in refluxing toluene also suggest
a slow diffusion of large amines. Ammonia and pyridine will be
quarternized by PFIEP to the extent of 95 and 57%, respectively.

The ester products from ethene (Equation 5) are methyl prOpion-

ate and methyl 4-oxo-hexanoate [19]. Conversions based upon ethene

are 44.1 and 2.4%, respectively. The origin of the latter compound
appears to be ethene insertion into the intermediate (L)nPdXCOCH2CH3.
1-Hexene yielded three products, methyl heptanoate, methyl 2-methy1-
hexanoate, and methyl 2-ethy1pentanoate in conversions of 13.7, 4.4,
and 0.7%, respectively (Equation 6). Cis-2-butene produced two pro-
ducts, methyl 2-methy1butyrate and methyl valerate in conversions
of 8.9 and 13.5%.
200 F. J. WALLER

Table 3. Effect of Olefin Structure on Carbomethoxylation Turnover

Rate with CF3S03H as Cocatalyst

Olefin acid a (mequiv) Conv.

ethene CF3S03H (2.27) 71.5 1688

l-hexene CF3S03H (2.27) 58.6 1362
propene CF3S03H (2.27) 47.4 828
cis-2-butene CF3S03H (2.29) 22.4 541

a(03P)4Pd (0.433 mmol Pd), 03P/(03P)4Pd 11.5, 500 mmol MeOH,

333 mmol olefin.

CH3(CH2)3CH=CH2 + co + CH30H ~ CH3(CH2)5C02CH3 +

f H3 f2 H5
CH3(CH2)3CHC02CH3 + CH3(CH2)2CHC02CH3 (6)

The most active catalyst is a palladium cation-exchanged PFIEP.

Under similar reaction conditions outlined in Table 1, a palladium-
exchanged PFIEP (1.34% Pd, 0.126 mmol Pd) carbomethoxy.lated propene
with a turnover rate of 1807.

A continuously stirred tank reactor (5 cc) charged with 0.90 g

of 30-60 mesh palladium-exchanged PFIEP (2.65% Pd) generated a pro-
duct stream of -35% by weight of butyrate esters when the CO, pro-
pene, and liquid composition [03P(0.9%)/MeOH(55.8%)/p-xylene] feed
rates were 9 g/h, 6 ml/h, and 2.4 ml/h, respectively. The catalyst
maintained constant activity over the length (12 hr) of the experi-


Palladium-catalyzed carbomethoxylation of olefins shows a rate

enhancement when cocatalysts, for example HCl or H2S04' are replaced
by homogeneous and heterogeneous perfluoroalkanesulfonic acids. In-
itial turnover rates gathered for a limited series of olefins suggest
that olefin coordination is the rate-determining step when non-
coordinating anions are present on the transition metal. In addi-
tion, the turnover activity for the same olefins is altered when
Nafion perfluorinated membrane is employed as the heterogeneous
acid. Possibly a slow diffusion rate within the fluorocarbon matrix
occurs when the olefin is larger than ethene. Continuous carbometh-

Table 4. Effect of Olefin Structure on Carbomethoxylation Turnover

Rate with PFIEP as Co catalyst

Olefin Acid a (mequiv) Conv.

Ethene PFIEP (2.26) 46.5 1065

Propene PFIEP (2.26) 23.1 719
l-Hexene PFIEP (2.26) 18.8 280

a(03P)4Pd (0.433 mmol Pd), 03P/(03P)4Pd 11.5, 500 mmol MeOH 333
mmol olefin.

oxylation with a palladium-exchanged Nafion suggests a carbonylation

catalyst possessing constant catalytic activity.


1. A. Mullen, in "New Syntheses with Carbon Monoxide", J. Falbe,

Springer-Verlag, New York, 1980.
2. K. Weissermel and Hans-Jurgen Arpe, "Industrial Organic Chem-
istry", Verlag Chemie, New York, 1978.
3. J. F. Knifton, J. Org. Chem., 41:2885 (1976).
4. (a) E. N. Squire and F. J. Waller, D. E. Offen. 2,739,096
(1978); assigned to E. I. du Pont de Nemours, Co.
(b) s. A. Butter, U.S. Patent 4,245,115 (1981); assigned to
Mobil Oil Corp.
5. E. N. Squire and F. J. Waller, U.S. Patent 4,292,437 (1981);
assigned to E. I. du Pont de Nemours, Co.
6. K. Bittler, N. von Kutepow, D. Neubauer and H. Reis, Angew.
Chem. Int. Ed., Engl., 7:329 (1968).
7. J. Tsuji, Acc. Chem. Res., 2:144 (1969).
8. D. M. Fenton, J. Org. Chem., 38:3192 (1973).
9. (a) G. Consiglio and M. Marchetti, Chimia, 30:26 (1976).
(b) G. Cavinato and L. Toniolo, Chimia, 33:286 (1979).
(c) G. Cavinato and L. Toniolo, J. Mol. Catal., 6:111 (1979).
(d) G. Cavinato and L. Toniolo, J. Mol. Catal., 10:161 (1981).
10. (a) S. A. Butter, U.S. Patent 3,700,706 (1972); assigned to
Mobil Oil Corp.
(b) M. Hara, K. Ohno, J. Tsuji, T. Kazimoto, S. Wakamatsu, and
R. Nakanishi, U.S. Patent 3,793,369 (1974); assigned to
Toray Ind., Inc.
11. R. Bardi, A. Del Pra, A. M. Piazzesi and L. Toniolo, Inorg.
Chim. Acta, 35:L345 (1979).
12. S. Enomoto, H. Wada, S. Nishita, M. Yanaka, and H. Takita,
202 F. J. WALLER

British Patent 1,597,814 (1981); assigned to Kureha Chem.

Ind. Co., Ltd
13. S. Otsuka, T. Yoshida, M. Matsumoto and K. Nakatsu, J. Am.
Chem. Soc., 98:5850 (1976).
14. H. Yoshida, N. Sugita, K. Kudo and Y. Takezaki, Bull. Chem.
Soc. Jpn., 49:2245 (1976).
15. G. W. Parshall, "Homogeneous Catalysis", John Wiley and Sons,
New York, 1980.
16. M. Hidai, M. Kokura and Y. Uchida, J. Organomet. Chem., 52:
431 (1973).
17. C. U. Pittman and Q. Y. Ng, U.S. Patent 4,258,206 (1981);
assigned to the University of Alabama.
18. A. Eisenberg and H. L. Yeager, ACS Symp. Ser., 180:1 (1982).
19. J. Tsuji, M. Morikawa and J. Kiji, Tetrahedron Lett., 1437
(1963) •


Clayton D. Wood and Philip E. Garrou*

Dow Chemical USA

Central Research - New England Laboratory
Wayland, MA 01778


Dicyclopentadiene (DCPD) is one of the major by-products of

industrial ethylene production via naphtha or gas oil cracking. One
potential upgrading process for DCPD is its hydroformylation to di-
cyclopentadienedimethanol (DCPDDM), which has found utility typical
of a difunctional alcohol in polyesters, epoxies, urethanes, etc.
The hydroformylation of this material is, however, not a trivial
process. Patent literature exists using either C02(CO)8 [1] or Rh203
[2]. The former suffers from high by-product formation, the need for
high (>3500 psig) pressures, and the need for a subsequent hydrogen-
ation step, while the latter, although giving excellent selectivity
(>95%), presents a difficult catalyst recovery situation in separat-
ing the high boiling product from the expensive homogeneous catalyst.
We wish to describe our studies on the hydroformylation of DCPD using
a C02(CO)8/PR3 catalyst system. a


Detailed analysis of the product profile during the reaction

reveals that olefin hydrogenation, DCPD cracking, aldol condensation,
and formate formation can compete effectively with the desired hydro-
formylation depending on such variables as temperature, pressure,
and the basicity of the PR3 modifier. Figure 1 shows the reaction

aWhile this work was in progress, a patent was issued on the hydro-
formylation of DCPD and similar materials using a Co2(CO)8/PR3
catalyst system [3].


J eO/R2

y No ~2

Figure 1. Dicyclopentadienedimethanol (DCPDDM) Formation Scheme

scheme for the production of diol. In the first step, the norbornyl
ring of the DCPD is hydroformylated to the aldehyde product. This
intermediate can either be reduced to the mono-alcohol or further
hydroformylated to the di-aldehyde intermediate. As the reaction
continues, both of these products are converted to diol.

The cobalt-phosphine catalyst systems are also known to be

active for the hydrogenation of olefins (Figure 2). Either ring can
be hydrogenated to give products that cannot be converted to the
diol. The high temperatures needed to hydroformylate DCPD can also
promote a reverse Diels-Alder reaction, which gives CPD (cyclopenta-
diene). Under the reaction conditions, CPD would be converted to
cyclopentylmethanol (Figure 3).

! eO/H 2

~O~ (}Q,OB
Figure 2. Olefin Hydrogenation Scheme

Figure 3. Cyclopentylmethanol (CPM) Formation Scheme

Figure 4 shows the six characterized formate products that are

derived from the aforementioned reaction mixture. Such products can
be formed via carbonylation of the aldehyde intermediates (see for-
mate section).

Dependence on Phosphine

Among the earliest studies of organophosphine modified cobalt

carbonyls as hydroformylation catalysts were those of Slaugh and
Mullineaux [4]. They described the use of alkyl-phosphine complexes
of cobalt carbony1s as catalysts to produce high yields of alcohols
relative to aldehydes at low pressures (500 psig). These results
were in contrast to those of cobalt carbonyl itself which requires
two reaction stages at high pressure to yield aldehydes in the first
step and then alcohols.

The major differences in hydroformy1ation activity between mod-

ified cobalt catalysts and C02(CO)S are in the areas of (1) reactiv-
ity, (2) hydrogenation activity, (3) catalyst stability, and (4)
ratio of linear to branched products. The first three of these cri-
teria are pertinent to the DCPD hydroformy1ation system. The reac-
tion rate of normal olefin hydroformy1ation has been shown to in-
crease with decreasing basicity of the phosphine ligand, although all
phosphine modified catalysts are slower than unmodified C02(CO)S [5].
This was explained by examination of the following equilibrium [6,7]:

~ OH

~ -CH ~ OCH
~II H6~1Io

8 0

Figure 4. Characterized Formate Products


co + HCo(CO)3L t HCo(CO)4 + L.

The equilibrium was shown to lie further to the left for more basic
and less bulky phosphines. Since the hydroformy1ation reaction is
much faster with HCo(CO)4 than with its phosphine complexes, faster
reaction rates should be obtained with triary1phosphines than with
the more basic tria1ky1phosphines. Intimately tied to the phosphine
effect on rate is its effect on selectivity, or more precisely the
hydrogenation activity, of the catalyst. The rate of reduction of
aldehyde has been shown to increase with increasing basicity of PR3
because compounds of the type HCo(CO)3PR3 are better reducing agents
than HCo(CO)4 due to their increased hydridic character [5-7]. Using
C02(CO)S as a catalyst in the 150-1S0°C temperature range, synthesis
pressures of 1500-4500 psig are necessary to prevent catalyst decom-
position. However, when C02(CO)S has been modified by triorganophos-
phines, pressures as low as 100-300 psig and temperatures as high as
150-200°C may be used without any observable catalyst decomposition.
Catalyst stability has also been shown to increase with the basicity
of the organophosphorous ligand [3].

Table 1 and Figures 5-7 reveal data for the hydroformy1ation of

DCPD using such C02(CO)S/PR3 catalyst systems. Our activity patterns
follow the normal trend: unmodified> P(ary1)3 > P(a1ky1)3' Reaction
of DCPD with unsubstituted C02(CO)S revealed the fastest reaction
rate and a yield of DCPDDM equivalent to that obtained with the best
of the phosphine systems studied but suffered due to catalyst decom-
position problems (discussed later). The unmodified cobalt catalyst
readily hydroformy1ates DCPD to the mono- and dia1dehyde products at
130-140°C (Figure 5). Higher temperatures are needed to reduce the
aldehyde to give DCPDDM.

The PPh 3 modified system revealed the next fastest rate, good
stability, and an equivalent DCPDDM yield. In the C02(CO)S-PPh3
system, DCPD reacts with synthesis gas at 160°C. As the reaction
temperature increased to 195°C, both the dia1dehyde and mono-alcohol
intermediates were formed. Hydroformy1ation of the 5-membered ring
proceeds at the same rate as the reduction of the mono-aldehyde in-
termediate. These products are then converted to DCPDDM.

With PPh3 we begin to see signs of higher amounts of hydrogena-

tion products (DCPD mono-o1s) and cracking products (cyclopentylmeth-
ano1). Pentapheny1phosphole (PPP) [S] was examined as a ligand,
since Shell [9] has reported that such ligands impart less hydrogena-
tion activity to hydroformylation systems than trialkylphosphines.
Using PPP, we observe the expected decrease in hydrogenation products,
a decrease in cracking products, but again no increase in DCPDDM se-
lectivity. The more basic tria1ky1phosphines [P(n-Bu)3 and P(Me)3]
revealed the expected slower rates but also unexpectedly only low
yields of diol. From analysis of the data in Table 1, it is concluded
that diol selectivity suffered due to the buildup of cracking and

Table 1. Dicyclopentadiene Hydroformylation with cobalt-organophos-

phine Catalysts

Cracking Products

Phosphine Condo pKa ova (Y'O? 0

1 a
2 c
3 PPP a b 2
4 PPh3 a 2.7 7
5 PPh3 c 2.7 4
6 PPh3 f 2.7 3
7 PPh3 g 2.7 1
8 P(n-Buh c 8.4 10 3
9 PMe3 a 8.6 20 5
10 PMe3 c 8.6 9 6
n P(C6Hnh a 9.7 3
12 P(C6Hn)3 c 9.7 7 1
13 P(C6 Hn)3 d 9.7 9
14 P(C6Hn)3 e 9.7 4 4

a4.0 g DCPD, 0.75 wt% Co, 195-200 o C, 2000 psig, H2/CO = 2, 3-4 hr
reaction time, P/Co = 1, 20 g solvent, 100% DCPD conversion
bUnknown although l-methylphosphole has pKa of +0.5
c5X amounts in footnote a
dSame as footnote a, except 500 psig H2/CO
eSame as footnote a, except 3500 psig CO/H2
fSame as footnote a, except 1.75 wt% Co
gSame as footnote c, except slow DCPD feed at operating conditions

hydrogenation products. The increase in cracking products can be

explained by the slower rate making more time available for the re-
verse Diels-Alder reaction to occur before all the DCPD has undergone
at least mono-hydroformylation. Comparing Figures 6 and 7, we ob-
serve the C02(CO)8/P(n-Bu)3 reaction reveals larger concentrations
of mono-ol-ene and mono-al-ene well past the one hr point of the re-
action, whereas much lesser amounts are observed in the PPh3 case.
The reduction to the mono-alcohol product is faster than the hydro-
formylation of the 5-membered ring in the P(n-Bu)3 system and very
little dialdehyde is noted. The last system studied is the P(C6Hll)3
modified system, which is an example of a bulky, basic (pKa = 9.7)
phosphine. In this case, one observes a decrease in cracking prod-
ucts due to a faster rate but selectivity to hydrogenation products
similar to that obtained for the trialkylphosphines. In spite of

Table 1- (Continued)

H~dro8enation Products Diol Products

H'\{}() HC~
+ +
Number 00 ~OH ~OCH

1 6 1 65 9
2 5 1 59 13
3 8 62 7
4 15 63 7
5 9 64 4
6 13 66 5
7 11 65 6
8 1 17 8 34 20
9 1 13 3 38 15
10 2 12 6 38 22
11 1 19 1 60 7
12 1 23 1 55 7
13 21 22 18 1
14 18 1 55 8

the 20% hydrogenation products, one obtains a DCPDDM yield close to

that of the unmodified C02(CO)8 or PPh3 modified systems. This can
be explained by looking at the mass balance data for these systems
in Table 2.

Gel permeation chromatography (GPC) of the reaction mixtures

indicated that higher molecular weight products, presumably aldol
condensation products, were produced. The selectivity of these prod-
ucts depended on the phosphine co-catalyst and decreased with in-
creasing basicity of the ligand. Table 2 shows that the combined
analytical data now account for all the products in the hydroformyl-
ation reaction.

In order to determine the stability of the catalysts, some of

the solutions were analyzed for cobalt concentration. The samples
were collected under reaction conditions. As noted in Table 3, the
PPh3 system was the most stable with 100% of the cobalt remaining
solubilized. The conventional C02(CO)8 catalyst was not as stable
even at higher reaction pressures (>3500 psig) when compared to the
cobalt-phosphine systems at 2000 psig. When the cobalt recovery was



o 2
Figure 5. Reaction profile for the hydroformylation of DCPD with

low, a gray, insoluble powder was isolated from the reaction mixtures.
Elemental analysis indicated that this product contained >95% cobalt.


The amount of hydrogenated product obtained was not surprising

since the literature reported that alkanes could be produced in as
high as 20% yield from terminal olefins using cobalt-phosphine sys-
tems [4]. However, the amount of formates formed is high compared
to the literature (only a few percent reported for terminal olefins).
The formation of formates during hydroformylation is a little studied
reaction [10]. Early in this research, a strong relationship between
phosphine identity and formate production was found. These results,
however, were not reproducible. For example, C02(CO)8/2PBu3 cata-
lyst systems produced between 11 and 31% formate products. The
C02(CO)8 purity, heat up schedule, catalyst concentration, oxygen,
water and a number of other "adventitious" variables were examined
and eliminated as causes of the irreproducibility. However, peroxide
contamination of the dicyclopentadiene has been identified as the






o 2


Figure 6. Reaction profile for the hydroformylation of DCPD with


source of the irreproducibility. When the olefin has been scrupu-

lously cleaned, one can reproducibly obtain data similar to that
shown in Table 1. Although the more basic PMe3 and P(n-Bu)3 ligands
produce more formates than the reactions promoted by PPh3, it was
found that P(C6Hll)3, which is the most basic ligand, showed a se-
lectivity similar to PPh3 (pKa's listed in Table 1). Our data is in
better agreement with a relationship between the stericbulk of the
phosphine ligands and the selectivity to formates; the phosphine
ligands with larger cone angles producing less formates. This trend
is also apparent when l-hexene is hydroformylated under similar con-
ditions with a variety of PR3 ligands [11].

Effect of Reaction Variables

Temperature. The reactor temperature is very important since

several side-reactions occur. The cobalt-phosphine systems need
temperatures >160°C to hydroformylate DCPD. At 160°C, the reduction
of aldehyde to alcohol is very slow so that higher amounts of aldol-


o 2

Figure 7. Reaction profile for the hydroformylation of DCPD with
C02 (CO)8-(n-Bu)3P

condensation products are obtained. At temperatures >180 o C, the

production of alcohol proceeds readily, but these conditions also
promote DCPD cracking.

Since different temperatures promote different side-reactions,

the reactor heat-up rate is also critical. The cobalt-PPh3 system
is the most stable to the effects of the reactor heat-up rate. For
the n-Bu3P modified catalyst, a quick heat-up rate gave 31% DCPDDM
in 2.5 hr, but a slow heat-up rate gave only 7% DCPDDM in the same
time. The distribution of products, as well as the stability of the
catalyst, are also affected.

Pressure. For the conventional cobalt catalyst, considerably

more catalyst decomposition was noted at 2000 psig, than at 3500
psig. This is consistent with the literature results for the C02(CO)8
catalyzed hydroformylation of other olefins [5-7].

The effect of total pressure on the product distribution for the

cobalt-P(C6Hll)3 system is reported in Table 4. Increasing the pres-
sure from 2000 to 3500 psig did not alter the results, but lower
pressures (500 psig) showed slower reaction rates, more hydrogenated
products, and less formate formation. The change in pressure did not
appear to effect the stability of the catalyst.

Table 2. Mass Balance Determination for the Hydroformylation of

DCPD with a Cobalt-Phosphine Catalyst

GLC Analysis GPC Analysis Mass

Phosphine Total % Products Total % Products Balance

81 24 105
84 15 99
87 14 101
83 21 104
92 11 103
P(C6H11)3 90 7 97
n-Bu3P 93 2 95
PPP 19 20 99
78 25 103

Table 3. Cobalt Analysis of Samples Isolated Under Reaction


Phosphine PPM Cobalt PPM Coablt % Co

(Before Reaction) (After Reaction) Recovered

1100 820 75
1400 1200 86
1100 1100 100
2500 2500 100

Reaction Poisons

The addition of water to the reaction mixture had no effect on

the results of the hydroformylation. Oxygen did not appear to be a
problem since the 3lp NMR data always showed excess PR3 ligand in the
mixture after the reaction, even in the case of P(n-Bu)3, which is
very sensitive to oxygen. In cases where DCPD cracking was a major
problem, (n5-C5H5)Co(CO)2 was isolated as the major organometallic
product from the reaction mixtures. This suggests that the CPD
formed from the DCPD cracking reaction is a potential poison for the

The catalysts were generally prepared in situ by the addition
of 2 equivalents of PR3 to C02(CO)8. The PR3 ligand readily reacts

Table 4. Effect of Pressure on the Cobalt-P(C6Hll)3 Catalyzed

Hydroformylation of DCPD

Pressure % Cracking % Hydrogenation % Formate % DCPDDM

3500 4 18 8 60
2000 3 19 8 60
500 9 43 <1 l8 a

a22% . rE)r) observed; reaction was not complete.


with C02(CO)8 to give yellow [Co(CO)3(PR3)2]+[Co(CO)4]- and red

[Co(CO)3PR3]2 [12,13]. The amount of each product depends on solvent
and reaction temperature. For example, in a non-polar solvent,
[Co(CO)3(PR3)2]+[Co(CO)4]- is converted to the red dimer at elevated
temperatures. Each product was isolated and used as a catalyst pre-
cursor for the hydroformy1ation of DCPD. The P(C6H11)3 system was
studied in detail. Figure 8 shows the IR spectra for [Co(CO)3P-
(C6H1l)3]2 and [Co(CO)3[P(C6H11)3]2+[CO(CO)4]-, respectively. The
IR and 31p NMR results are summarized in Table 5.

The reaction mixture from the coba1t-P(C6Hll)3 catalyzed hydro-

formy1ation of DCPD was isolated at room temperature under nitrogen
and analyzed by IR and 31p NMR spectroscopy. The 31p NMR spectrum of
the reaction mixture showed four resonances in the 70-82 ppm range
(Figure 9). These same four resonances are observed regardless of
the catalyst ~recursor - C02(CO)8 and 2P(C6Hl1)3 in situ, {[Co(CO)3-
[P(C6 H11)3J2} [Co(CO)4]-, or [Co(CO)3P(C6H11)3]2. The signals at
79.2 and 75.9 ppm correspond to [Co(CO)3P(C6H11)3J2 and {Co(CO)3[P-
(C6Hl1)3J2}+[CO(CO)4J-, respectively. The IR spectrum of the reac-
tion mixture (Figure 10) confirms that these products are present and
suggests that HCo(CO)3P(C6H11)3 and C02(CO)7P(C6H11)3 are the other
two products.

An authentic sample of C02(CO)7P(C6H11)3 was prepared by treat-

ing a mixture of C02(CO)8 and [Co(CO)3P(C6H11)3]2 with CO. An IR
spectrum of this compound showed three bands at 2079 (m), 2020 (m),
and 1988 (s) cm-1 , which were also observed in the IR spectrum of the
reaction mixture. The 31p NMR analysis confirmed that C02(CO)7-
P(C6Hll)3 was a product in the reaction mixture (Table 5).

The IR band at 2038 cm-1 suggests the presence of HCo(CO)3P-

(C6Hl1)3 in the reaction mixture. In order to confirm this, a mix-


.I I I I I I I I
2100 2000 1900 1800 2100 ·2000 1900 1100

Figure 8. IR spectra of (A) [co(CO)~P(C6Hll)3]2 in toluene,

(B) [Co(CO)3[P(C6Hll)3]2] [Co(CO)4]- in THF

ture of C02(CO)8 and 2 equivalents of P(C6Hll)3 was treated with

synthesis gas at elevated pressure and temperature. 3lp NMR analysis
(Figure 11) of the reaction mixture at -50°C indicated the presence
of C02(CO)7P(C6Hll)3 and [CO(CO)3P(C6Hll)3]2, and a signal was also
present that was assigned as HCo(CO)3P(C6Hll)3. The proton c0upled
3lp NMR experiment caused these signals to broaden due to the cou-
pling of the phosphorous nuclei to the hydrogen atoms on the cyclo-
hexyl ring. The signal at 72.8 ppm was also coupled to another hy-

Table 5. Summary of the IR and NMR Data for the Organometallic

Compound Isolated from the DCPD Hydroformylation with a
Cobalt-P(C6Hll)3 Catalyst

Compound 3lp NMR (ppm)

Z079 (m)
ZOZO (m)
1987 (s)
1963 (m)
1943 (s)
[Co(CO)3[P(C6Hll)3]Z]+ 75.9 1995 (m)
[Co(CO)4]- 1980 (m)
1890 (s)
ZO.l Z038 (m)
1953 (s)

drogen atom (Jp_H = 47 Hz). This coupling is consistent with a

H-Co-P interaction. A selective lH decoupled 3lp NMR spectrum elim-
inated the P-cyclohexyl coupling and revealed the cobalt-hydride

o CO2(CO)7 PR3
A [Co(CO)3PRJ2
• [Co(CO)3(PR3)2 + J
o <> HCo(CO)3PR3

82 76 70 PPM

Figure 9. 3lp{lH} NMR spectrum of a DCPD hydroformylation reaction

mixture containing CoZ(CO)8-P(C6Hll)3 catalyst reaction

2100 2000 1900 1800 (em-I)

Figure 10. IR spectrum of the reaction mixture from the hydro-

formylation of DCPD with C02(CO)8-P(C6Hll)3

coupling (Figure 11). The sharpness of this peak was temperature

dependent and became broader at room temperature. This fluxional be-
havior accounted for the broad resonance observed in Figure 9 since
this spectrum was taken at room temperature. The lH NMR data were
also consistent with the reported results for HCo(CO)3PR3 [14]. A
doublet is observed at 20.17 with JH-P = 49 Hz (Figure 11).





Figure 11. 31p NMR spectra of the reaction product of C02(CO)8/2

P(C6H11)3 with CO/H2. Spectra obtained at -50°C: (Top)
proton decoup1ed and (Bottom) selective decoup1ing of
cyc10hexy1 protons.

These results indicate that four organOmetallic complexes can

be characterized from the DCPD hydroformy1ation with a cobalt-phos-
phine catalyst, i.e. [Co(CO)3(PR3)2] [Co(CO)4] , [Co(CO)3PR3]2, C02-
(CO)7PR3, and HCo(CO)3PR3. These data agree with the results report-
ed for a C02(CO)8/n-Bu3P system that was studied by in situ IR spec-
troscopy [15]. The in situ study also indicated that HCo(CO)4 is
present in such systems. Lack of stability under the conditions used
to study our reaction mixtures precluded observation of HCo(CO)4 in
the current studies.


Examination of the product profile during the hydroformy1ation

of dicyc10pentadiene using C02(CO)8/PR3 catalyst systems revealed
olefin hydrogenation, DCPD cracking, aldol condensation, and formate
formation can compete effectively with the desired hydroformy1ation,
depending on such variables as temperature, pressure and the nature
of the PR3 modifier. Because of the aforementioned competitive re-
actions, attempts to optimize the yield of DCPDDM have not resulted
in any better yield than one obtains with C02(CO)8 itself, although
much more stable catalyst systems have been obtained.


Unless indicated otherwise, all operations were conducted under

purified argon or nitrogen, using standard inert atmosphere tech-

Infrared spectra were determined on a Beckman IR 4240 spectrom-

eter. NMR spectra were recorded on a JEOL FX90-Q spectrometer
equipped with a broad-band, tunable probe. Phosphorous-3l chemical
shifts were referenced to external 85% H3P04' GLC analyses were per-
formed on Hewlett-Packard 5730A gas chromatograph using a 30 m 8E-30
capillary column. A detailed description of the product analysis
will be given elsewhere [16]. HPLC analyses were performed on a
° styra-gel and two 100 A
Waters ALC-GPC 201 equipped with one 500 A °
styra-gel columns in series. Cobalt analyses were carried out via
the wet-ash method or with a 8MI IV plasma emission spectrometer.

DCPD Hydroformylation

In a typical reaction, a mixture of 4.0 g DCPD, 1.0 g dodecane

(internal standard), 0.08 g C02(CO)8, 1 or 2 equivalents of PR3 and
20 ml solvent (toluene, THF, toluene/THF (3/1) mixture, or hexane)
was loaded into a glass reactor vessel in an inert atmosphere drybox.
The vessel was placed into a 300 cc stirring reactor (Autoclave En-
gineers) and the system was purged with argon. The reactor was pres-
surized to 1600 psig of synthesis gas and heated to 195-200°C in
about 20 min. At that temperature, the reactor pressure was -2000
psig. Reaction times ranged from 2-4 hr. The reactor was then
cooled to room temperature and depressurized to give a red solution
which was analyzed.

Catalyst Preparations

Preparation of [CO(CO)3(PR3)Z]+[CO(CO)4]-. These compounds were

prepared as described in the literature [12-13]. A toluene solution
containing 0.50 g C02(CO)8 was treated with 0.82 g P(C6Hll)3 and
stirred for 1 hr. Filtering the mixture gave 0.90 g of yellow Co-
(CO)3[P(C6Hll)3]2+[CO(CO)4]- (70% yield).

Preparation of [CO(CO)3PR3]2' These products were prepared as

described in the literature [12-13]. A toluene solution of 0.50 g
C02(CO)8 and 0.82 P(C6Hll)3 was heated to reflux under nitrogen for
2 hrs. The toluene was concentrated in vacuo and 1.1 g of red
[Co(CO)3P(C6Hll)3]2 was isolated by filtration (90% yield).

Preparation of C02(CO)7P(C6Hll)3. A mixture of 0.10 g [Co(CO)3-

P(C6Hll)3]2 and 0.6 g C02(CO)8 in 10 ml of toluene was heated to 60°C
for 4 hr under CO. A red solution was isolated and IR analysis con-
firmed the presence of C02(CO)7P(C6Hll)3 [17].

Characterization of HCo(CO)3P(C6H")3. A mixture of 0.20 g

C02(CO)8 and 0.32 g P(C6Hll)3 in 20 ml toluene was heated in the
stirring autoclave at 200°C for 0.5 hr under 2000 psig of synthesis
gas. The reactor was cooled to O°C via an ice bath and the contents
were isolated. The toluene was removed in vacuo at O°C. The red
residue was dissolved into 5 ml of d8-toluene and cooled to -78°C to
remove the dimer and salt products. The solution was then analyzed
by lH and 3lp NMR spectroscopy at -50°C.

Isolation of (n5-CsHs)(Co(CO)2. A DCPD hydroformylation reac-

tion was carried out in which DCPD trimer and tetramer were the major
products, and a product was concentrated and distilled at 2 mm pres-
sure. Several drops of a red liquid was isolated at 38°C. The IR
bands at 2030 (s) and 1970 (s) cm- l and the boiling point are con-
sistent with those for (n5-C5H5)Co(CO)2 [18].


1. K. BUchner, O. Roelen, and J. Meis, U.S. Patent 2,850,536

(1958); assigned to Ruhrchemie.
2. J. Falbe, Canadian Patent 893,716 (1972); assigned to Ruhrchemie.
3. H. Shibatani, Y. Ogomori, T. Kameda, and Y. Yanagi, GB Patent
2,046,251 (1980); assigned to Mitsubishi Petrochemical Co.
4. L. H. Slaugh and R. D. Mullineaux, J. Organomet. Chem., 13:469
(1968) •
5. F. E. Paulik, Catal. Rev., 6:49 (1972).
6. E. R. Tucci, Ind. Eng. Chern., Prod. Res. Dev., 9:516 (1970).
7. E. R. Tucci, Ind. Eng. Chem., Prod. Res Dev., 7:125 (1968).
8. F. C. Leavitt and F. Johnson, U.S. Patent 3,412,119 (1968);
assigned to Dow Chemical Co.
9. Shell International Research, G. B. Patent 1,110,549 (1968).
10. I. Wender and P. Pino, "Organic Synthesis via Meta1-Carbony1s,"
John Wiley and Sons, New York, 12 (1977).
11. C. D. Wood and P. E. Garrou, to be published.
12. W. Heiber and W. Freyer, Chem. Ber., 91:1230 (1958).
13. O. Voh1er, Chem. Ber., 91:1235 (1958).
14. W. Heiber and H. Duchatsch, Chem. Ber., 98:2933 (1965).
15. M. van Boven, N. Alemdarog1u, and J. M. L. Penninger, J. Organ-
omet. Chem., 84:65 (1975).
16. R. A. Dubois, C. D. Wood, and P. E. Garrou, to be published.
17. P. Szabo, L. Fekete, G. Bor, Z. Nagy-Majos, and L. Marko,
J. Organomet. Chem., 12:245 (1968).
18. R. B. King, "Organometallic Synthesis," Academic Press, New
York, 119 (1965).

S. Suzuki, J. B. Wilkes, R. G. Wall, and S. J. Lapporte

Chevron Research Company

Richmond, California 94802


The manufacture of chemicals from synthesis gas is of consid-

erable interest, as synthesis gas can be made from a wide variety
of carbon-containing feedstocks including coal, coke, petroleum,
natural gas, and biomass. Methanol and acetic acid are now being
made commercially from synthesis gas and carbon monoxide, respective-
ly. A commercial plant for the production of acetic anhydride from
methanol and carbon monoxide is now under construction.

The synthesis of ethylene glycol, as shown in equation 1, is

potentially one of the most economically attractive chemical uses of

2CO(g) + 3H2(g) + HOCH2CH20H(l) (1)

synthesis gas since all of the feed appears in the product. There
is no loss of feed components to form water in this reaction, such
as occurs in the production of ethanol, olefins, or aromatics from
synthesis gas.

Sherwin [1] of Chem Systems, Inc., has identified three synthe-

sis gas-based routes to ethylene glycol that are being developed.
The apparently simplest route is the direct reaction of synthesis gas
to form ethylene glycol. Another route is the oxidative dimerization
of carbon monoxide in the presence of methanol to form dimethyl ox-
alate, followed by hydrogenation of the oxalate esters to ethylene
glycol. The third route involves reacting formaldehyde, carbon mon-
oxide, and water to form glycolic acid, followed by esterification
of the glycolic acid and hydrogenation of the ester to ethylene
glycol. Sherwin indicates that the glycolic acid process to ethylene


glycol shows a good technical prognosis for success; but that to

provide a 20% cost advantage over the ethylene-based process, a drop
in the price of methanol relative to naphtha is required. The Chem
Systems economic estimates are, of course, based upon information
disclosed in patents. Chevron Research engineering cost estimates,
based upon additional research and development data, show more fa-
vorable economics.

Each of the three routes from synthesis gas to ethylene glycol

has certain advantages and problems. The one-step production of
ethylene glycol from synthesis gas appears attractive, as it involves
only a single reaction step and requires no separation of the synthe-
sis gas components. This process has been the subject of consider-
able research. The Gibbs free energy of the reaction shown in equa-
tion 1 is -12 kca1/mo1e at 25°C, indicating that this reaction is
thermodynamically very favorable at moderate temperatures. However,
the formation of ethylene glycol from synthesis gas with the tran-
sition metal catalysts that have been used up to the present time
has required temperatures in the vicinity of 225°C to achieve rea-
sonable reaction rates. At this temperature, the Gibbs free energy
of reaction is about +17 kca1/mo1e, which is highly unfavorable.
Consequently, very high pressures are required to obtain useful con-
versions to ethylene glycol. In addition, other reactions occur si-
multaneously. Methanol is a major by-product. It has been proposed
[2] that the low reactivity of the transition metal catalysts may be
due to the formation of formaldehyde from carbon monoxide and hydro-
gen as a necessary intermediate in equation 1 and to the low concen-
trations of formaldehyde that are thermodynamically attainable by
the direct reaction of carbon monoxide and hydrogen. The existence
of free formaldehyde as an intermediate has been disputed [3], but
an intermediate formyl complex, which may have similar thermodynamic
limitations, has been proposed.

While a one-step synthesis of ethylene glycol from synthesis

gas is an attractive concept, experience has shown that commercial
chemicals are often produced more economically by multiple-step re-
actions, where each reaction step can be carried out at conditions
that are favorable thermodynamically and kinetically. The production
of formaldehyde is an excellent example of this approach. The direct
synthesis of formaldehyde from synthesis gas in significant yields
is thermodynamically impossible. However, formaldehyde is produced
in high yields from methanol by oxidative dehydrogenation at ele-
vated temperature and low pressure, while the methanol feed is made
in high yields from synthesis gas by catalytic reaction at moderate
temperatures and pressures. The overall result is the production
of formaldehyde from synthesis gas in high yield by a two-step proc-
ess, which avoids the thermodynamic limitations for formation of
formaldehyde. Similarly, ethylene glycol can be produced from syn-
thesis gas in good yields by production of oxalic or glycolic acid
esters, followed by hydrogenation of the esters to ethylene glycol.

The coupling of carbon monoxide with palladium catalyst in the

presence of alcohols and nitric acid or alkyl nitrites has been de-
scribed by Nishimura and coworkers [4]. The commercialization of
oxalic acid production, presumably by this process, has been an-
nounced. This oxalic acid process reportedly gives high yields, but
requires complete separation of carbon monoxide from hydrogen, and
entails the production and recycle of nitrogen oxides and alkyl ni-
trites. Problems might be expected to arise in this hydrogenation,
as oxalic acid is more readily decarboxylated than malonic acid in
ethylene glycol solution [5] at similar reaction conditions. How-
ever, a recent patent on the hydrogenation of the oxalate esters [6]
to ethylene glycol shows good rates and yields.

The Chevron route to ethylene glycol from synthesis gas is also

a multiple-step process, which uses the reaction of formaldehyde
with carbon monoxide to form glycolic acid, or polyglycolides, as
shown in simplified form in equation 2. The glycolic acid, or poly-

CH20 + H20 + CO ~ HOCH2COOH (2)

glycolides, are then reacted with an alcohol to form a glycolate

ester, which is catalytically hydrogenated to ethylene glycol, as
shown in equation 3. The production of ethylene glycol via glycolic

HOCH2COOR + 2H2 ~ HOCH2CH20H + ROH (3)

acid was practiced commercially at one time by the Du Pont Company.

However, the Du Pont process of producing glycolic acid required
unusually high pressures and operated at conditions giving high cor-
rosion rates with conventional materials of construction. These
problems have been overcome by the process developed at Chevron Re-
search Company, which is described in this paper.



The thermodynamic data for glycolic acid formation are both in-
complete and unreliable. However, the magnitudes of enthalpy and
entropy changes are large enough to provide useful information. The
reaction, as written in equation 2, is exothermic with ~Ho298 = -27
± 3 kcal/mol.

Assuming the enthalpy of reaction is -27 kcal/mol, it is esti-

mated that ~SO = -59 cal/oC. Ignoring heat capacities, and with the
drastic assumption of ideal solutions, the Gibbs free energy and
equilibrium constants can be estimated and equilibrium concentrations
and pressures can be calculated usi~g the equation:

K=------- (4)

(Pco is in atmospheres, and the activities of other compounds in

equation 4 are in mole fractions in the liquid phase.) Results of
these calculations for 98% and 95% formaldehyde conversions are shown
in Figure 1. It can be seen that very high pressures are required
for good conversions at temperatures above 200°C, as used in the Du
Pont process, but that the reaction should be essentially irrevers-
ible at pressures of carbon monoxide of 10-50 psig at temperatures
below 100°C. The entropy of reaction given above was estimated by
assuming that the heat of solution of formaldehyde vapor is the same
as the heat of condensation of methanol vapor, that the entropy of
liquid formaldehyde is similar to that of liquid methanol or formic
acid, and that the entropy of glycolic acid is the average of the
entropies of liquid acetic and lactic acids. The uncertainty in
these entropy estimates is about ± 6 cal/oC, which gives an uncer-
tainty of an additional ± 3 kcal/mol in the Gibbs free energy f~r the
reaction. These uncertainties prevent exact predictions of equilib-
ria, but the high enthalpy of reaction is the main contributing
factor to the free energy, thus permitting a reasonable estimate of
carbon monoxide pressures versus formaldehyde conversions in ideal
solutions. Strong acids and oxygenated compounds are known to form
solutions that deviate greatly from ideality, but our experience
leads us to believe that the deviations in the activities of the re-
actants must compensate for each other in such a manner as to permit
the estimated carbon monoxide pressures in Figure 1 to be fairly
close to the actual equilibrium pressures.

High Pressure Process

The reaction of formaldehyde, water, and carbon monoxide to form

glycolic acid was discovered by Loder [7]. His examples show the
use of BF3' and of sulfuric, hydrochloric, phosphoric, or formic
acids as catalysts. A 94% yield of glycolic acid was obtained at
temperatures of l30-200°C and a pressure of 900 atmospheres. The
commercial process is, perhaps, best described in Cody's patent, [8]
on the purification of glycolic acid. According to this patent, gly-
colic acid is synthesized by the carbony1ation of formaldehyde in a
solvent of water and glycolic acid, with 0.9-1.5 wt % sulfuric acid
as the catalyst, at pressures of about 7000 psig and temperatures of
about 220°C. Purification of the dark colored crude product includes
carbon treatment and removal of sulfuric acid and trace metals by
ion exchange. According to Shattuck [9], the conversion of formal-
dehyde to methyl formate by the Cannizzaro reaction can cause sub-
stantial losses in yield in this process. This yield loss is mini-
mized by maintaining a low concentration of formaldehyde through the
use of multiple reactors and by adding the formaldehyde in stages.

I BallI
~H" =-27 keel
.MS0 =-59 CaI.I"C
Ideal Solutlonl
Feed: One Mole H20
Per Mole CH20

Temperature, °C

Figure 1. Estimated carbon monoxide pressures; glycolic acid from

formaldehyde and water

The low concentration of formaldehyde maintained in this manner

presumably favors carbony1ation over the Cannizzaro reaction, which
is bimolecular in formaldehyde.

The high pressure used in the Du Pont glycolic acid process ob-
viously requires use of unconventional equipment and probably causes
high energy use in the process. In addition, sulfuric acid at 220°C
is highly corrosive, and conventional corrosion-resistant materials
of construction cannot be used at these conditions.

Metal-Catalyzed Carbony1ations

Subsequent to the early work by Loder [7] and by Shattuck [9],

Bhattacharyya and coworkers [10] investigated the synthesis of gly-
colic acid from formaldehyde, carbon monoxide, and water using
nickel, cobalt, and iron halides as the catalysts. Nickel iodide
was the best catalyst, giving 42% conversion of formaldehyde to gly-
colic acid at 200°C and 8700 psi in three hours. They investigated
other temperatures and pressures and obtained about 10% conversion
of formaldehyde to glycolic acid at 1500 psi at 200°C.

The formation of glycolic acid by transition-metal halides was

explored at Chevron Research Company [11] as part of an investiga-
tion of carbony1ation reactions. Yields of 45-72% glycolic acid in
the form of the acetate ester were obtained by reacting the formal-
dehyde trimer, trioxane (1,3,5-trioxacyc10hexane), and acetic acid
for five hours at temperatures of 125-150°C and a pressure of 1000
psig with catalysts of Rh(CO)2C1 and iodide. HBr and HI alone were
less active catalysts than rhodium iodide but were considerably
better catalysts than HC1 or H2S04 [12]. A yield of 52% glycolic
acid as the acetate ester was obtained with HBr catalyst in five
hours at 150°C and 1000 psig compared to 12% and 23% yields with the
same molar amounts of HC1 or H2S04 as catalysts. This work demon-
strated that reasonable conversions of formaldehyde to glycolic acid
could be obtained by the carbony1ation of formaldehyde at moderate
temperatures and pressure, but the reactions were still slower than
desired. In addition, the bromides and iodides are corrosive, and
the use of acetic acid solvent added to the complexity of product

Low Pressure Glycolic Process

The discovery of the use of concentrated hydrogen fluoride as

a catalyst for the reaction of formaldehyde with carbon monoxide [13]
provided a greatly improved route to the production of glycolic acid.
With hydrogen fluoride catalyst, the carbony1ation can be carried
out at reasonable rates at temperatures lower than O°C and pressures
as low as 10 psig, although somewhat higher temperatures and pres-
sures are economically more attractive. It is possible to use syn-
thesis gas as the source of carbon monoxide in this reaction and,
in effect, to use the carbony1ation reaction to separate carbon mon-
oxide from the hydrogen in the synthesis gas. With HF catalyst, the
Cannizzaro reaction of formaldehyde is negligible at normal reaction
conditions, so that it is not necessary to minimize formaldehyde
concentrations or to stage the addition of formaldehyde when carbon-
y1ating formaldehyde to glycolic acid. Conventional corrosion-re-
sistant metals can be used as materials of construction. The hydro-
gen fluoride can be completely removed from the products by vola-
tilization [14].

Hydrogen fluoride is a very active catalyst for the carbony1a-

tion of formaldehyde, but additives such as HBF4 or cuprous or silver
salts can increase the activity of the catalyst [15]. The addition
of HBF4 increases the activity of the catalyst significantly but
leads to difficulties in removing the products from the catalyst.
The presence of cuprous salt in the HF catalyst caused an initially
rapid reaction of carbon monoxide but resulted in little increase in
the rate of formation of glycolic acid. The additional amount of
carbon monoxide consumed in the carbony1ation of formaldehyde with
HF and Cu+ was consistent with the formation of [Cu(CO)3]+. The
rapid formation of cuprous tricarbonyl appears to account for the

initially increased rate of reaction of carbon monoxide. Copper(I)

carbonyl in HZO/BF3 or concentrated sulfuric acid has been reported
by Suoma and Sano [16] to increase the rate of carbonylation of
formaldehyde to glycolic acid. The maximum concentration of form-
aldehyde used in their experiments was 1.3 wt %, and the minimum
useful concentration of sulfuric acid was 84.5%. These results are
interesting theoretically but offer little promise of application to
a commercial process.

Hydrogen fluoride is used extensively in the petroleum industry

for alkylation. According to UOP [17], more than 90 commercial HF
alkylation units of their design were in operation, under design,
or in construction in 1980. Other companies also design this type
of plant; so the number of HF alkylation units being used commercial-
ly is undoubtedly considerably higher. In spite of this widespread
commercial use of hydrogen fluoride, hydrogen fluoride tends to be
a rarity in chemical laboratories, probably because of the special
precautions required in handling this material. Some of the tech-
niques of handling of hydrogen fluoride, as well as the properties
of HF, have been reviewed by Dove and Clifford [18].

HF-catalyzed carbonylations of formaldehyde to glycolic acid

have been carried out at Chevron Research Company in batch auto-
claves, in a continuous stirred-tank reactor, and in a small pilot
plant. Results obtained in all of these units were in good agree-
ment. The experiments reported here will be restricted to the batch
tests, as these covered a wider range of variables.

The formation of glycolic acid, as shown in equation Z, requires

water as a reactant. With a deficiency of water, diglycolic acid is
formed, as shown in equation 5:


The deficiency of water leading to formation of diglycolic acid is

not necessarily a stoichiometric deficiency for formation of glycolic
acid but is a deficiency in terms of the reaction equilibria and ki-
netics, as discussed below.

A plausible reaction mechanism is shown in Scheme 1. Formal-

dehyde is present in aqueous solution principally in the form of the
hydrate, methylene glycol 1, and a series of oligomers [19]. Pro-
tonation and dehydration of methylene glycol can lead to the carbon-
ium ion Z. This reaction is favored in concentrated HF solution by
the high-activity of hydrogen ion and the low activity of water.
Addition of carbon monoxide and water leads to the formation of gly-
colic acid 4. Glycolic acid readily undergoes self-esterification
to form oligomeric hydroxyesters or polyglycolides, even in aqueous
solution. Polyglycolide 5 may also be formed by the reaction of the
acyl cation ~ with glycolic acid, as shown in Scheme 1.

Glycolic Acid Formation

-----<.~HOCH2+ + H20

HOCH 2+ + CO -----<.~HOCH2CO+
2 3

-----<.~HOCH2COOH + H+

- - - - - - t... HOCH2C02CH2COOH + H+
3 4 5

Diglycolic Acid Formation

HOCH 20H + HOCH 2COOH ~ ..

..~===+ HOCH 20CH 2COOH + H20
4 6

HOCH 20CH 2COOH + H+ - - - -..~ +CH 20CH 2COOH + H20

6 7

Scheme 1. Postulated mechanism of HF-catalyzed carbonylation of


The mechanism of formation of diglycolic acid is also shown in

Scheme 1. Formaldehyde can react with glycolic acid to give the
hemiacetal §. Protonation and dehydration of this hemiacetal gives
the carbonium ion 7, which reacts with carbon monoxide and water to
give diglycolic acid §. Diglycolic acid could also presumably be
formed by the acid-catalyzed reaction of two moles of glycolic acid
to form water and diglycolic acid in a typical acid-catalyzed forma-
tion of an ether from an alcohol, but tests have shown that this
reaction is not significant at our reaction conditions. Hemiacetals
such as 2 are readily formed from formaldehyde and alcohols; and the
amount of diglycolic acid formed should be a function of the activ-
ities of the reactants, especially water, and of the relative con-
centrations of methylene glycol and glycolic acid hemiacetal 2' The
product distribution will also depend on the relative reaction rates
of these intermediates to give glycolic and diglycolic acids. In
the absence of water, methylene glycol and hemiacetals will not be
present, but formaldehyde will be present in polymeric form and the
CH20H groups, shown in Scheme 1, will be present as -CH20CH20-
groups, and these entl.ties may react to give diglycolic acid or its

The reaction mechanism may also be written with steps similar

to those shown in Scheme 1 but without the assumption of the forma-
tion of carbonium ions and acyl cations either in the free state or
as ion pairs. Protonated compounds such as HOCHZOH2+ are undoubt-
edly present, and this protonated methylene glycol may add carbon
monoxide with displacement of water without the existence of a free
carbonium ion. Similarly, the acyl cation HOCHZCO+ 2 shown in Scheme
1 may be stabilized as the acyl fluroide, HOCHZCOF. From the ther-
modynamic properties of the pure compounds, we estimate the Gibbs
free energy of the reaction


to be between +6 and +9 kcal. These values give estimated equilib-

rium constants of 10- 5 to 10- 7 for reaction 6. However, the strong
tendency of HF to form FHF- compounds [20] may significantly stabi-
lize the acyl fluoride in concentrated HF solution as the complex
HOCH2COF-HF, which could be an intermediate in the carbonylation re-
actions. We have no direct evidence for the formation of the acyl
fluoride-HF complex, but this complex is analogous to the (CH3)3CCO-
Cl-SbC1 5 complex found by Nojima et al. [21] to be an intermediate in
the reaction of (CH3)3CCI with carbon monoxide in liquid SbC15-S02
solution. The analogous stabilization of HOCHZCO+ in concentrated
HBr, HI, or H2S04 solution would not be expected; and the existence
of a stabilized acyl fluoride complex may account, in part, for the
unusually high activity of HF in catalyzing carbonylation of formal-
dehyde to glycolic acid.

We have found [2Z] that alkoxyacetic acids and their esters are
readily formed by the carbonylation of formaldehyde or methylal with
HF catalyst in the presence of alcohols. This reaction is so favor-
able that any methanol present in the formaldehyde used for the pro-
duction of glycolic acid is converted almost quantitatively into
methoxyacetic acid. Diglycolic acid can be hydrogenated to diethyl-
ene glycol, but methoxyacetic acid is an undesirable by-product, as
it cannot readily be converted to a glycol. To minimize the pro-
duction of methoxyacetic acid, formaldehyde with a low methanol con-
~ent is required for commercial production of glycolic acid and
ethylene glycol by this process.

The formation of methoxyacetic acid probably proceeds by a mech-

anism very similar to that shown in Scheme 1. Methanol reacts read-
ily with formaldehyde to form the hemiacetal, CH30CH20H, and methyl-
aI, CH30CH20CH3' The reaction of alcohols with formaldehyde to form
hemiacetals is thermodynamically favored over the formation of
methylene glycol. The equilibrium constant for hemiacetal formation
is about 30 for methanol and 16 to 19 for other primary alcohols
[23]. The acid-catalyzed elimination of water from a hemiacetal,
ROCH20H, in a manner similar to that shown in Scheme 1, will give
the carbonium ion ROCH2+ and lead to alkoxyacetic acid instead of

glycolic acid. With HF catalyst, carbonylations in alcoholic solu-

tions are faster than carbonylations of aqueous formaldehyde at sim-
ilar conditions, probably because the ROCH2+ carbonium-ion interme-
diates are stabilized by the electron-donating alkoxy groups. This
combination of favorable equilibria for hemiacetal formation combined
with high reaction rate results in a high selectivity for the conver-
sion of methanol in formaldehyde to methoxyacetic acid.


Diglycolic acid is a valuable material. It can be converted

to diethylene glycol, or it can be isolated and used as a monomer,
surfactant builder, or for other typical uses of dicarboxylic acids.
In most of our work, the desired material was glycolic acid; and re-
action conditions were selected to minimize the production of digly-
colic acid.

The results given in Table 1 show some of the factors influenc-

ing the relative amounts of glycolic and diglycolic acids formed in
the HF-catalyzed carbonylation of formaldehyde. The ratio of water
to formaldehyde in the reaction mixture is obviously the predominant
factor, with the fraction of formaldehyde converted to diglycolic
acid increasing from 5% with 0.5 moles of water per mole of formal-
dehyde to 51% of the formaldehyde converted to diglycolic acid in
the absence of water but with otherwise similar reaction conditions.
This is consistent with the mechanism shown in Scheme 1. Increasing
the reaction temperatures resulted in the formation of somewhat
higher ratios of diglycolic to glycolic acid in the products. The
effect of carbon monoxide pressure on diglycolic acid formation was
not investigated; but it is believed that pressure is not a signifi-
cant factor, as subsequent work at total reaction pressures as low
as 50 psig did not reveal any diglycolic acid formation that was not
consistent with the results given in Table 1.


Analysis of the experimental results shows that the rate of

carbonylation of formaldehyde with HF catalyst is first order in
formaldehyde concentration and approximately first order relative to
carbon monoxide pressure. An increase of 4 mole % in HF concentra-
tion results in a doubling of the rate of carbonylation. The range
of rates that are possible in this reaction within practical oper-
ating conditions is unusually large. Rates can be changed about 10-
fold by a 40°C change in temperature, about 50-fold by changing HF
concentration from 60-84%, and 10-fold by changing carbon monoxide
pressure from 40-400 psi. The cumulative effect is a possible 5000-
fold change in rate of conversion of formaldehyde within this range
of conditions. The large effect of HF concentration on rate permits

Table 1. Dig1yco1ic Acid Formation in Formaldehyde Carbony1ation "T1
products, mol %
run conditions s::m
starting of CH20 charged -I
material reaction :::I:
test mole ratio, time, glycolic dig1ycolic unreacted >
temp., °c pressure, psig
no. CH20/H20/HF min acid acid CH20, % or
1 1/0.5/2.5. 50 2070-1050 80 91 5 Ul
2 1/0.5/2.5 1
53 1980 50 77 6
3 1/0.5/2.5 70 2000 35 90 10 -1 :::I:
4 1/0.2/2.5 50 1900 35 67 23 Ul
5 1/0.0/2.5 25 1940 45 34 45 >5 enG)
6 1/0.0/2.5 50 1920 120 28 51 8 >
7 1/0.0/1.5 50 1950 155 24 55 >3 Ul


reasonable rates to be obtained at low carbon monoxide pressures and

extremely rapid rates to be obtained at high hydrogen fluoride con-
centrations and moderate carbon monoxide pressures.

Some variables affecting the rates of formaldehyde conversion

in this process are shown in Figure 2. The conversions of formal-
dehyde in these experiments were calculated from the amount of carbon
monoxide consumed versus reaction time and the unreacted formaldehyde
remaining at the end of each experiment. The first order nature of
the reaction in formaldehyde is apparent from the linear plots of
the logarithm of the amounts of unconverted formaldehyde versus re-
action time in Figure 2. Tests 19 and 20 show the effects of tem-
perature upon reaction rate at similar pressures. Additional data
on these tests is given in Table 2. Rates which are rapid enough to
be of commercial interest are obtainable at pressures of 80-340 psig
at a temperature of 50°C, as shown in Tests 12 and 19 (Figure 2 and
Table 2). With high carbon monoxide pressure, the rate is rapid,
even with low concentrations of HF (Test 2).

The factor with greatest effect on reaction rate is the concen-

tration of hydrogen fluoride in the reaction mixture. For example,
at 68 wt % HF, SO°C, and 80 psig (Test 16), the reaction half life
time is almost three times greater than with 81 wt % HF at the same
temperature and pressure (Test 12). The effects of HF concentration
on reaction rates are consistent with a rate that is proportional to
the ability of HF to protonate the methylene glycol form of formalde-
hyde, or an analogous derivative, as shown in Scheme 1. The ability
of an acid to protonate an organic molecule can be roughly measured
by the Hammett acidity function, -Ro, which is essentially [log(aH+)
+ constant] for any single proton acceptor. If the rate of carbonyl-
ation of formaldehyde is related to the ability of HF to protonate
formaldehyde, the rate at two different concentrations of HF should
be proportional to 10~Ro, where ~Ro is the difference of acidity
functions at two different concentrations of HF. Unfortunately, He
is not a uniquely defined function,as Ho depends on the solvent and
on the nature of the indicators used in its determination. In addi-
tion, two determinations of the Hammett acid function of aqueous HF
[24,25] gave considerably different values for Ro, as shown in
Figure 3. For our purposes, we have assumed that the more recent
values by Dallinga et al. [25] are correct, especially as their re-
sults do not show the anomalous sharp deviations from linearity of
the dependence of the acidity function on concentrations that were
found by Hyman et al. [24].

Within the rather restricted range of concentrations that are

of practical interest, we have found that the rate of carbonylation
can be related to Dallinga's values for -Ro. The correlation of the
effect of acidity function of HF solutions on the rates of acid-cat-
alyzed carbonylations of formaldehyde does not appear to be directly
applicable to other protonic acids. The limited information avail-




HF, Temp., Press.,

Test Wt % ~ pSig
2 53 53 1980
12 81 50 80
16 68 50 80
19 60 50 340
20 60 30 355


20 40 60 80 100 120
Reaction Time, Min.

Figure 2. Effects of temperature, pressure, and HF concentration on

rates of formaldehyde conversion at test conditions shown
in Tables 1 and 2.

able shows that concentrated sulfuric acid is a less active catalyst

than HF for formaldehyde carbony1ation. In addition, Kurkov et a1.
[12] showed that HBr was a considerably more active catalyst than an
equimo1ar amount of sulfuric acid for the carbony1ation of formalde-
hyde in acetic acid, in spite of the fact [26] that sulfuric acid
is a stronger acid than HBr in equimo1ar concentrations in water,
as measured by the acidity functions. The unexpectedly low activity
of sulfuric acid compared to HF, HBr, and HI as catalyst for the
carbony1ation of formaldehyde may be related to the limitations of
acidity functions in predicting the protonation of oxygenated com-
pounds. Arnett and Scorrano [27] point out that the original Ho
functions were generated chiefly from the reactions of primary amines
in dilute solution, but that different classes of bases follow dif-
ferent acidity functions. In addition, in concentrated solutions
the organic base itself may contribute a serious medium effect to its
own ionization. An alternative explanation is that acyl halides,
especially acyl fluorides, may be intermediates in the reaction mech-
anism, as previously discussed.

Table 2. Glycolic Acid from Formaldehyde, Carbon Monoxide, and Water

starting reaction reaction time, products, mol %

materials conditions minutes of CH20 charged

test mole ratio wt % temp. , pressure, to constant glycolic dig1ycolic unreacted

No. CH20/H20/HF HF °c psig pressure acid acid CH20, %

8 1/1.8/16.7 84 22 970-830 11 50 83
9 1/1.8/16.7 84 22 500-355 9 60 87
10 1/1.4/12.3 82 23 1000-655 10 47 97
11 1/1/10 81 0 50 147 a 73 27 b
12 1/1/10 81 50 80 80 122a 60 b 0.0
13 1/1/8.3 78 23 960-700 60 150 93
14 1/1/5 68 37 1920-1600 30 50 105 b
15 1/1/5 68 20 50 207 52 44 b
16 1/1/5 68 50 80 60 46 37 b
17 1/1/3.6 60 50 1500 38 212 91 1.0 1
18 1/1/3.6 60 50 1660-1500 25 40 95 3.5
19 1/1/3.6 60 50 340 110 104b
20 1/1/3.6 60 30 355 110 67 21b
21 1/0.5/2.5 56 50 2490-2135 35 88 2.7
- ---------------------- - - - - c
aSma11 reactor with limited stirring. N
bAna1ysis approximate in these experiments. ~


c 6

• Hyman, et al., 1957

• Dallinga, et al., 1970
60 80
__ ~-- __

Hydrogen Fluoride, Mole %

Figure 3. Acidity function of aqueous hydrogen fluoride

Carbonylation Techniques

Most of our carbonylation reactions were run in a 300-ml 316 SS

magnetically stirred autoclave. Some runs at low pressures were
conducted in a 205-ml metal vessel about 35 mm ID and 150 mm high
which was stirred by a magnetic stirring bar at the bottom of the
reactor. This vessel was equipped with an ice water cooled reflux
condenser to avoid loss of HF in some of the low pressure runs.

Paraformaldehyde was used as the formaldehyde source in some

preliminary product studies and in some reaction rate studies. Tri-
oxane was used as the formaldehyde source in most runs, as it is a
pure compound with reproducible properties. Aqueous formaldehyde in
the concentrations required is not available commercially but must
be made at the plant site, as concentrated formaldehyde solutions
must be kept hot to keep the formaldehyde in solution, and these so-
lutions slowly decompose on standing. Paraformaldehyde is a variable
material containing from 91-98% CH20 with the remainder being mainly
water. The variable nature of paraformaldehyde makes it a poor ma-
terial for studying formaldehyde carbonylations.

In early runs, HF was measured volumetrically in a polypropyl-

ene graduated cylinder and added to the open autoclave. Later, HF
was transferred through Teflon lines to a stainless steel gas sample
cylinder resting on a scale, weighed, and pressured through Teflon
lines into the autoclave. Full-body protective clothing was used
during HF handling. The final product mix was transferred to a
polypropylene or Kel-F vessel by suction or by pipette. Kel-F ves-
sels are preferred, as polypropylene contains antioxidants and organ-
ic materials, which can be extracted into the products [2B].

The analytical procedure was to first vacuum strip HF under

water aspiration to an autoclave temperature of BO-95°C (nitrogen
stripping in Kel-F was also occasionally used to minimize corrosion.)
The crude product was thermally esterified by reacting 0.25-0.5 grams
with 7 grams of methanol in a l4-ml 316 SS "micro bomb" at 200°C for
30 minutes. A variety of GLC columns may be used for analysis. A
lIB-inch x 5-foot FFAP column was satisfactory, with 2-butanol as
internal standard. Unreacted formaldehyde was determined as methylal
in this procedure, while the acids were determined as their methyl
esters. Residual HF in the final product was determined by fluoride
ion electrode.


Glycolic acid in aqueous solution can be hydrogenated directly

to ethylene glycol with ruthenium catalysts [29]. The yields did
not exceed B3% in the published work, and pressures above 9000 psi
were usually used. Our tests with ruthenium catalysts and aqueous
glycolic acid showed similar yields at lower pressures; but the re-
action mixture was quite corrosive at reaction conditions and metals
dissolved by corrosion were deposited on the ruthenium, resulting in
rapid changes in selectivity. In more recent work [30], mixed pal-
ladium-rhenium catalysts were used to hydrogenate glycolic acid in
solution in dioxane or tetrahydrofuran with somewhat greater success.
Yields of ethylene glycol varying from less than BO% up to 93.B%
were reported. However, expensive solvents appear to be required;
and reaction rates were slow relative to the amounts of metal cata-
lyst required. Slow reaction rates appear to be characteristic of
rhenium catalysts used for the production of alcohols from acids and
discourage efforts to commercialize use df these catalysts.

The hydrogenation of esters is the classical route from carbox-

ylic acids to alcohols and still appears to be the most attractive
process. Hydrogenation of esters to alcohols tends to be equilibrium
limited at the usual reaction conditions. High conversions are fa-
vored by low reaction temperatures and high hydrogen pressures. For
the hydrogenation of glycol glycolate in glycol (equation 7), we es-

HOCH2CH200CCH20H(l) + 2H2(g) + 2HOCH2CH20H(I) (7)


timate that ~H~98 = -19.9 ± 1 kcal/mol and ~S~98 = -47.4 cal/oK.

This thermodynamic estimate is based upon the literature and on our
experience that at 225°C and 100 atm pressure about 99% of a 50 wt %
solution of glycolate ester can be converted to ethylene glycol.
This is more favorable thermodynamically by about 2 kcal than the
literature values for hydrogenation of ethyl acetate to ethyl alcohol
and Adkins' results on the equilibrium in the hydrogenation of
n-octyl caprylate [31]. A probable explanation of this discrepancy
is that ethylene glycol is more strongly hydrogen bonded than a
simple alcohol such as ethanol. Compounds such as ethylene glycol
and 2-methoxyethanol can form five-membered hydrogen-bonded rings.
The extra enthalpy for this ring formation is about 2 ± 0.7 kcal/mol

Du Pont, at one time, made ethylene glycol from glycolic acid.

Larson [33] described the vapor-phase hydrogenation of methyl gly-
colate and other glycolate esters with various copper-containing
catalysts. Conditions used were typically 3 vol % ester in hydrogen
at 200-225°C and about 450 psig pressure. Cockerill [34] patented
a method of vaporizing glycolate esters into hydrogen in a falling
film type of vaporizer in a manner that avoided oligomerization or
decomposition of the glycolic esters in order to avoid carrying the
decomposition products into the hydrogenation reactor. Loder [35]
patented the liquid-phase hydrogenation of glycolic esters at pres-
sures in excess of about 6000 psi. It has generally been assumed
that the vapor-phase hydrogenation of methyl glycolate to ethylene
glycol was used commercially, but there is little information to sub-
stantiate this assumption.

The methyl glycolate-ethylene glycol system is not very suitable

for a vapor-phase reaction. At 210°C, the saturation vapor pressure
of ethylene glycol is 19.3 psi; while the partial pressure of 3 mol
% ethylene glycol at 440 psia is 13.2 psi, which is 68% of the glycol
saturation vapor pressure at 210°C. From the Kelvin equation [36],
it can be calculated that catalyst pores less than 4-8 nm in diam-
eter should be filled with liquid ethylene glycol at these condi-
tions, if the glycol surface tension is in the region of 24-48
dynes/cm. Consequently, most of the pores of a typical catalyst will
be filled with liquid at 210°C and 440 psia, and will be unavailable
for a vapor-phase reaction. In addition, the recovery of the prod-
ucts from the dilute stream of organic vapors in hydrogen will be

These considerations caused us to direct our efforts toward de-

velopment of the hydrogenation of glycolic esters in trickle bed re-
actors. Trickle bed reactors are now widely used commercially, es-
pecially in the petroleum industry. With the glycolic ester in the
liquid phase in a trickle bed reactor, there is no requirement for
a volatile ester, such as methyl glycolate, and no need to use low
hydrogen pressure to maintain vapor-phase operations.

In choosing the alcohol part of the ester to be used in the

glycolic ester hydrogenation, methanol appears advantageous because
of its low cost relative to other alcohols. However, the low boil-
ing point of methanol poses a number of problems in the process.
The vapor pressure of pur~ methanol is about 600-900 psi in the 200-
220°C region where the hydrogenation is conducted. A liquid-phase
hydrogenation of methyl glycolate would require relatively high pres-
sures due to the partial pressure of methanol liberated in the hydro-
genation reaction. In preparing the methyl ester from the glycolic
acid/polyglyco1ide product, it is not possible to drive the esteri-
fication to completion by distilling the water produced in the es-
terification away from the partially esterified product. The ester-
ification procedure is relatively complicated, as complete conversion
of the glycolic acid to the ester is impossible except by separating
the methyl glycolate from the unconverted glycolic acid and recy-
cling or by removing a mixed methanol-water stream, separating the
methanol from the water, and recycling the dry methanol. Esteri-
fication of the glycolic acid without addition of esterification cat-
alysts is desirable to avoid added impurities. This type of esteri-
fication requires temperatures of 150°C or above for reasonable re-
action rates, and the high vapor pressure of the methanol at these
temperatures requires costly high pressure reactors.

Primary alcohols with boiling points above 100°C are consider-

ably simpler to use in the process, as these alcohols permit distil-
ling or azeotroping the water of esterification from the esterifica-
tion products, and the vapor pressures of the alcohols are not ex-
cessive at reaction conditions. Based on cost and availability, we
have used 2-methyl-I-propanol, ethylene glycol, and diethylene
glycol as the alcohols to prepare glycolate esters for hydrogenation
studies. Ethylene glycol and diethylene glycol are produced in the
process, and these alcohols have been used in most of our work. In
laboratory esterifications with ethylene glycol or diethylene glycol,
the esterifications were carried out at 205-220°C pot temperature
while distilling out the water of reaction through a 20-tray 01der-
shaw column. The esterifications were 99+% complete in 45-70 min-

Due to the pioneering work of Homer Adkins [31], copper chro--

mite is the catalyst most commonly chosen for the hydrogenation of
esters to alcohols or glycols. Copper chromite is predominantly
cupric oxide and cupric or cuprous chromite, CuO + CuO·Cr203 or
CuO + CU20·Cr203. The active reduced catalyst can be regarded as
highly dispersed, supported copper metal [37]. There seems to be
no reason to believe that copper chromite catalyst is inherently su-
perior to other dispersed or supported copper catalysts. Nickel and
cobalt have also been used for the hydrogenation of esters to alco-
hols, but these catalysts tend to give lower yields of alcohols than
those obtained with copper-containing catalysts.

The commercial copper chromite catalysts that were available

at the time this investigation was started were found to be too low
in activity for the economical hydrogenation of glycolic esters.
Our investigations led to catalysts with substantially improved ac-
tivity and stability for glycolic ester hydrogenation. The first
generation of these improved catalysts [38] was made by coprecipi-
tating copper, cobalt, and zinc hydroxy-carbonates and converting
them to the oxides by calcining. It was subsequently found [39] that
addition of 5-25% colloidal silica to the hydroxy-carbonates before
calcining gave catalysts of higher activity, and this improved cata-
lytic activity was accompanied by greater surface area and pore
volume and substantially decreased bulk density of the catalysts.
The preparation, testing, and performance of these catalysts is de-
scribed in considerable detail in the patents [38,39] and will not be
repeated here. Table 3 summarizes some of the hydrogenation results
obtained in a small, continuous hydrogenation unit using feed con-
taining 47-52% ethylene glycol glycolate in ethylene glycol. The
catalysts used in these small-scale, continuous hydrogenations were
crushed and screened to 14-28 mesh, as our catalyst beds were 16-17
cm long (excluding preheater section); and the studies of Montagna
and Shah [40] showed that catalyst particles larger than 14 mesh
result in inadequate catalyst wetting and liquid holdup in trickle
bed reactors less than 50 cm in length. When possible, reaction con-
ditions were selected which gave 85-96% conversions of the glycolate
esters. At higher conversions, differences in degree of reaction
were difficult to establish by analysis. In addition, the kinetics
for the conversions of the last traces of glycolate ester indicated
that diffusion limitations or bypassing problems developed at low
glycolate concentrations in the small laboratory continuous reactor.

An example of glycolate hydrogenation by a supported copper

catalyst, which was not designed specifically for liquid-phase hy-
drogenation, is shown in Table 3, Item 1. An alumina-stabilized,
copper-zinc oxide synthesis gas shift catalyst was used in this ex-
periment. A commercial manganese-promoted copper chromite catalyst
(Item 2), which is one of the new high activity copper chromites, was
considerably more active--but less so than our cobalt-copper-zinc
oxide catalyst. The addition of colloidal silica during the prepa-
ration of a catalyst of this type improved the activity enough (Item
5) so that it was necessary to reduce the reaction temperature by
about 24° in order to keep the conversion of glycolic ester at the
desired level. With the proper amount of added colloidal silica, it
was possible to obtain a copper-zinc oxide catalyst (Item 6) that
was more active than even the silica-modified cobalt-copper-zinc
oxide catalysts.

Selectivity in catalytic hydrogenation is an important consid-

eration. Very small amounts of methanol and ethanol have been formed
in all of our hydrogenations. High temperatures and excessive re-
action times result in the hydrogenolysis of ethylene glycol with

Table 3. Catalytic Hydrogenation of Ethylene Glycol Glycolate

properties reaction
catalyst components
pelleted catalysts a conditions c

type, surface pore bulk ester

mole silica, temp. ,
item as area, volume, densit~,b LHSV conversion,
ratio wt % °c
oxides m2 /g cm 3 /g g/cm mole %

1 Cu-Zn-Al 33/65/2 none 25-40 d 1.3 215 1.6 46

2 Cu-Cr-Mn 38/31/3 none 59 0.16 1.2 218 3.2 89
3 Co-Cu-Zn 1/1/1 none 91 0.14 1.7 218 3.2 95
4 Co-Cu-Zn 1/1/1 5 140 0.29 1.1 193 3.2 78
5 Co-Cu-Zn 1/1/1 15 179 0.38 0.9 194 3.2 94
6 Cu-Zn 2/1 12 113 0.39 0.8 182 3.2 96

aproperties in unreduced oxide form.

bBulk density of 14-28 mesh crushed tablets.
cAll tests at 1500 psig pressure. C/)

dSurface as specified by manufacturer (United Catalysts). C/)


formation of considerable additional ethanol and other materials,

with but little additional formation of methanol. We speculate that
most of the methanol arises from decarboxylation of small amounts
of glycolic acid present in the feed. Our selectivities to ethyl-
ene glycol from the hydrogenation of glycolic esters are usually
above 95% of theory. This high selectivity makes the exact determi-
nation of yields quite difficult. The yields can, undoubtedly, be
influenced by catalyst modifications. One highly active copper
chromite catalyst has been tested that gave ethylene glycol yields
of less than 90% of theory, even at mild conditions. While this is
an undesirable result, it does demonstrate that catalyst modifica-
tions can significantly affect the yields of ethylene glycol.


Investigation of the liquid-phase hydrogenation of glycolic

esters in trickle bed reactors led to the discovery that polymeric
glycolic esters can cause rapid catalyst deactivation [41]. As gly-"
colic acid contains both hydroxyl and carboxyl groups, polymeric
esters of glycolic acid will be present in glycolic esters and in
their solutions in alcohols when the alcohol-ester mixtures are al-
lowed to reach equilibrium. In the Chevron process for making gly-
colic acid, polyglycolide is formed during the process of stripping
off the hydrogen fluoride catalyst.


Esterification of this material starts with polymeric glycolic acid.

It is possible to esterify these polyglycolides to the extent that
essentially all the free carboxyl groups have been esterified without
reaching equilibrium in regard to the amount of polymeric esters
present. Depolymerization can be effected by heating the alcohol-
ester mixtures for an additional period of time. The reaction pro-
ceeds as shown in equation 9, which uses the trimeric ester as an


example. It has been found that, in the absence of a catalyst, it

is necessary to heat the alcohol-ester mixture for 50-90 minutes at
190-230°C after completion of the initial esterification in order to
reach the equilibrium level of polymeric esters. The equilibrium
amount of polymeric ester and the degree of polymerization in the
mixtures will depend upon the concentration of the ester in the al-
cohol and the type of alcohol.

The amounts of polymeric esters and their effects on catalyst

deactivation were determined with a series of solutions of diethylene
glycol glycolate (DEGG) in diethylene glycol. These solutions were
made from glycolic acid, which had been extensively purified to

remove all of the usual catalyst poisons such as halides, metals,

and sulfur compounds. The results are summarized in Table 4 and in
Figure 4. The total amount of glycolic acid in these solutions in
the esterified form was determined from the saponification number,
and was calculated as monomeric DEGG to obtain the concentrations of
DEGG shown in Table 4. The distribution of glycolic ester monomer
and polymers was determined by acetylating samples of solution with
acetic anhydride and pyridine and analyzing the acetylated product
by gas chromatography using a flame ionization detector. A relative-
ly short gas chromatographic column was used, e.g. a 3-mm x l30-cm
OV-lOl column. The amounts of each type of ester shown in Table 4
are based upon area percents, as pure reference compounds were not
available. The esters designated as dimer and trimer should be a
mixture of isomers in the case of these diethylene glycol esters as
the analytical method will not distinguish between, for example,
glycolic dimer esterified with one hydroxyl of a diethylene glycol
molecule and diethylene glycol esterified on both hydroxyls with
glycolic acid.

The high stability of the catalyst when hydrogenating a feed

with a low content of polymeric glycolic ester is shown in the upper
part of Figure 4. The ester used for this hydrogenation is that used
in Test 29C shown in Table 4. This test proceded for approximately
1000 hours at various reaction conditions with no measurable loss of
catalyst activity after an initial period for stabilization.

The rapid loss of catalyst activity when hydrogenating a feed

with a high content of polymeric glycolic ester is shown in the lower
part of Figure 4, which shows the change in catalyst activity with
time during a continuation of Test 28 (Table 4). The rapid loss of
activity is apparent. Increasing reaction temperatures intermit-
tently to increase catalyst activity was effective in increasing
conversions for only a few hours.

The catalyst deactivation is related to the polymeric glycolic

ester in the feed and not to content of total glycolic ester, as can
be seen by comparing Tests 21 and 29 in Table 4. For Test 21, a DEGG
with a concentration of 56% ester was made, diluted with an equal
weight of diethylene glycol, and then hydrogenated without transes-
terification. This feed retained a high content of polymeric esters,
and a rapid deactivation of the catalyst resulted. Depolymerization
of the glycolic esters, as discussed above (Test 29C) with a similar
total ester concentration, gave a feed that did not deactivate the

The exact degree of glycolic polymerization that is permissible

in these liquid-phase hydrogenations has not been established. Very
little catalyst deactivation occurred in Test 29B (Table 4), in which
the feed contained some trimer esters and a fairly high level of
dimer esters. Tentatively, we consider that the feed should contain

Table 4. Hydrogenation of Diethylene Glycol Glycolate: Effect

of Polymeric Esters on Catalyst Stability


28 29A 29B 29C

feed, wt % DEGGa 59 61 39 30 28

ester composition, %
monomer 59 66 78 89 74
dimer 33 31 21 11 22
trimer 8 3 0.7 none 4

test conditions b
temp., °c 200 182 182 204 182
pressure, psig 1500 1500 1500 1500 1500
LHSV 4.8 4.0 4.8 6.4 3.2

test period, hr 98 19 107 86 22

conversion, %
initial 96.3 96.3 93.5 92.5 95.7
final 81.2 93.6 93.1 92.5 93.0

loss of activity
(% conversion)/100 hr 15 14 0.37 none 12

aDEGG = diethylene glycol glycolate.

bCata1yst 2:1 Cu:Zn oxides and 11.9% silica, except Run 21.
CCata1yst 1:1:1 Co:Cu:Zn oxides + 11.5% silica in Run 21.

less than 22% of the glycolic acid in the form of polymeric esters
in order to avoid catalyst deactivation.


Difunctiona1 compounds other than those described in the pre-

ceding sections of this paper can also be made by carbony1ating al-
dehydes other than formaldehyde or by reacting difunctiona1 .a1coho1s
along with aldehydes in HF-cata1yzed carbony1ations. Some of these
reactions are described below.

The HF-cata1yzed carbony1ation of formaldehyde in the presence

of ethylene glycol gives 1,4-dioxan-2-one (p-dioxanone), along with

Run 29C with Depolymerlzed Glycolate

1990C 1500 I ~204°C, 1000 pslg
, psg .... ~
. '. ,r~~~----~.r--w".

Run 28 with Polymeric Glycolate

, I

,, ,,,
I 204°C'
199°C I

;I. I

J , I

I , I

R••ctlon PNUUN, 1100 pslg

Ttlt Period, Hr
Figure 4. Effects of polymeric glycolate esters on catalyst

considerable amounts of glycolic and dig1yco1ic acids. The formation

of glycolic and dig1yco1ic acids in this reaction is a result of the
incorporation into the carbony1ation products of the water formed
when the lactone is produced from ethylene glycol, carbon monoxide,
and formaldehyde. The production of water during this reaction can
be avoided by carbony1ating dioxo1ane. The carbony1ation of 1,3-
dioxolane with HF catalyst [42] as shown in equasion 10 gave more
than 90% yield of 1,4-dioxan-2-one.


The 1,4-dioxan-2-one is an interesting monomer that can be polymer-

ized to give a material suitable for production of sutures [43].
The lactone can also by hydrogenated to diethylene glycol or dehy-
drogenated in the presence of a base [44] to give diglycolic acid.

The carbonylation of formaldehyde in the presence of diols other

than 1,2-diols yields additional hydroxy acids or dicarboxylic acids
[45]. Carboxylation of formaldehyde in the presence of 1,4-butanedi-
01 gave 3,8-dioxadecanedioic acid, HOOCCH20CH2CH2CH2CH20CH2COOH, and
4-hydroxybutoxyacetic acid, HOOCCH20CH2CH2CH2CH20H. The use of di-
ethylene glycol instead of 1,4-butanediol in this reaction gave the
analogous 3,6,9-trioxaundecanedioic acid and 8-hydroxy-3,6-dioxaoc-
tanoic acid.

The HF-catalyzed carbonyl at ion of acetaldehyde also provides a

simple route to the production of lactic acid from acetaldehyde [46].

CH3CHO + CO + H20 + CH3-CH-COOH (11)


The yield of lactic acid from acetaldehyde is considerably more sen-

sitive to reaction conditions than the yield of glycolic acid from
formaldehyde. We have not tried to optimize the reaction conditions.
A yield of 80 mole % lactic acid was obtained by conducting the car-
bonylation at room temperature with 2.5 times the theoretical amount
of water and by pumping the acetaldehyde into the reaction mixture
over a period of 20 minutes. Only 25 mole % yield of lactic acid was
obtained when 1.5 times the theoretical amount of water was used and
when all of the reactants were combined at the beginning of the

The thermodynamics of the carbonylation of acetaldehyde and

water to give lactic acid are known with considerably more accuracy
than the thermodynamics of formation of glycolic acid from formalde-
hyde, water, and carbon monoxide, although, as is the case with the
thermodynamics of glycolic acid formation, it is impossible to pre-
dict how closely the thermodynamics in liquid hydrogen fluoride ap-
proximate the thermodynamics calculated for pure compounds or for
pure compounds in dilute aqueous solution. We calculated that the
reaction of liquid acetaldehyde and water with carbon monoxide to
give aqueous lactic acid solution has a Gibbs free energy of -4.87
kcal/mol at 25°C and +5.97 kcal/mol at 200°C. The carbonylation of
acetaldehyde and water to give lactic acid is, therefore, thermody-
namically favored by low temperatures in a manner very similar to the
carbonylation of formaldehyde and water to glycolic acid.


Hydrogen fluoride is a very active catalyst for aldehyde carbon-


y1ations, and its use permits conducting these carbony1ations at low

to moderate temperatures. At these temperatures, the reactions are
thermodynamically favorable,. so that essentially complete conversion
of the aldehydes can be obtained at low pressures of carbon monoxide.
This high catalytic activity of hydrogen fluoride, along with the
feasibility of removing the hydrogen fluoride from the products by
volatilization, makes the use of hydrogen fluoride for catalytic con-
version of synthesis gas, alcohols, and aldehydes to chemicals eco-
nomically attractive.


1. M. B. Sherwin, Hydrocarbon Process., 60(3):79 (1981).

2. D. R. Fahey, J. Am. Chem. Soc., 103:136 (1981).
3. B. D. Dombek, ACS Symp. Ser., 152:213 (1981).
4. K. Nishimura, S. Uchiumi, K. Fujii, K. Nishihara, and H. Itat-
ani, Preprints, Div. Petro Chem., ACS, 24(1):355 (1979).
5. L. W. Clark, J. Phys. Chem., 71:2597 (1967).
6. L. R. Zehner and R. W. Lenton, U.S. Patent 4,112,245 (1978);
assigned to Atlantic Richfield Co.
7. D. J. Loder, U.S. Patent 2,152,852 (1939); assigned to E. I.
du Pont de Nemours & Co.
8. N. F. Cody, U.S. Patent 3,859,349 (1975); assigned to E. I. du
Pont de Nemours & Co.
9. M. T. Shattuck, U.S. Patent 2,443,482 (1948); assigned to E. I.
du Pont de Nemours & Co.
10. S. K. Bhattacharyya and D. Vir, Adv. Cata1., 9:625 (1957).
11. S. J. Lapporte and V. P. Kurkov, in "Organotransition-Meta1
Chemistry," Y. Ishii and M. Tsutsui, Eds •• Plenum Publishing
Corp., pp 199 (1975).
12. V. P. Kurkov. S. J. Lapporte, and W. G. Toland, U.S. Patent
3,801,627 (1974); assigned to Chevron Research Co.
13. S. Suzuki, U.S. Patent 3,911.003 (1975); assigned to Chevron
Research Co.
14. S. Suzuki, U.S. Patent 4,188,494 (1980); assigned to Chevron
Research Co.
15. S. Suzuki, U.S. Patent 4,016,208 (1977); assigned to Chevron
Research Co.
16. Y. Souma and H. Sano, Nippon Kagaku Kaishi, 263 (1982).
17. Hydrocarbon Process., 59(9):176 (1980).
18. M. F. A. Dove and A. F. Clifford, in "Chemistry in Anhydrous,
Prototropic Inorganic Solvents," Vol II, G. Jander, H. Span-
dau, and C. C. Addison, Eds., Pergamon Press (1971).
19. J. F. Walker, "Formaldehyde," 3rd Ed., Reinhold, pp 52-72
(1964) •
20. J. W. Larson and T. B. McMahon, J. Am. Chem. Soc., 104:5848
(1982) •
21. M. Nojima, F. Shiba, M. Yoshimura, and N. Tokura, Chem. Lett.,
1133 (1972).

22. S. Suzuki, U.S. Patent 3,948,977 (1976); assigned to Chevron

Research Co.
23. P. Le Henaff, C. R. Acad. Sci., Ser. C, 262:1667 (1966).
24. H. H. Hyman, M. Kilpatrick, and J. J. Katz, J. Am. Chern. Soc.,
79:3368 (1957).
25. G. Da11inga, J. Gaaf, and E. L. Mackor, Reel. Trav. Chim.
Pays-Bas, 89:1068 (1970).
26. M. A. Paul and F. A. Long, Chern. Rev., 57: 1 (1957).
27. E. M. Arnett and G. Scorrano, "Advances in Physical Organic
Chemistry," Vol 13, V. Gold and D. Bethell, Ed., Academic
Press, pp 88-90 (1976).
28. R. W. Dabeka, A. Mykytiuk, S. S. Berman, and D. S. Russell,
Anal. Chern., 48:1203 (1976).
29. J. E. Carnahan, T. A. Ford, W. F. Gresham, W. E. Grigsby, and
G. F. Hager, J. Am. Chern. Soc., 77:3768 (1955).
30. D. Freudenberger and F. Wunder, U.S. Patent 4,214,106 (1980);
assigned to Hoechst Aktiengese11schaft.
31. H. Adkins, "Organic Reactions," Vol VIII, R. Adams, Ed., Wiley,
pp 1-27 (1954).
32. P. J. Krueger and H. D. Mettee, J. Mol. Spectr., 18:131 (1965).
33. A. T. Larson, Brit. Patent 575,380 (1946); assigned to E. I.
du Pont de Nemours & Co.
34. R. F. Cockerill, U.S. Patent 2,316,564 (1943); assigned to E. I.
du Pont de Nemours & Co.
35. D. J. Loder, U.S. Patent 2,285,448 (1942); assigned to E. I. du
Pont de Nemours & Co.
36. J. R. Anderson, "Structure of Metallic Catalysts," Academic
Press, p 379 (1975).
37. G. Natta and R. Rigamonti, in "Handbuch der Kata1yse," Vol V,
G. M. Schwab, Ed., Springer-Verlag, Vienna, pp 569-573
38. R. G. Wall, U.S. Patent 4,113,662 (1978); assigned to Chevron
Research Co.
39. J. B. Wilkes, U.S. Patent 4,199,479 (1980); assigned to Chevron
Research Co.
40. A. A. Montagna and Y. T. Shah, Ind. Eng. Chern., Process Des.
Dev., 14:479 (1975).
41. J. B. Wilkes, U.S. Patent 4,366,333 (1982); assigned to Chevron
Research Co.
42. S. Suzuki, U.S. Patent 4,070,375 (1978); assigned to Chevron
Research Co.
43. N. Doddi, C. C. Versfe1t, and D. Wasserman, U.S. Patent
4,052,988(1977), assigned to Ethicon, Inc.
44. S. Suzuki, U.S. Patent 4,142,057 (1979); assigned to Chevron
Research Co.
45. S. Suzuki, U.S. Patent 4,057,577 (1977); assigned to Chevron
Research Co.
46. S. Suzuki, U.S. Patent 3,948,986 (1976); assigned to Chevron
Research Co.

Alan Peltzman

Halcon SD Technology
Two Park Avenue
New York, New York 10016


This paper describes the Scientific Design/Bethlehem Steel

process for producing formic acid by carbonylation of methanol.
The process, jointly developed by our companies, in effect con-
verts CO and water into formic acid and exhibits feedstock
flexibility as well as attractive economics.

We expect the demand for formic acid will greatly exceed

the capacity available from traditional byproduct routes, and
we look to our new technology to close this gap. It should be
noted that this technology has the outstanding feature that
feedstock synthesis gas composition variability can be readily

I will review the traditional routes to formic acid, discuss

its conventional uses and trace the development of the Scientific
Design/Bethlehem Steel process. I will also discuss the possible
sources of synthesis gas and its relation to the economics of
formic acid via our technology and describe our process. Econo-
mics will then be presented for different designs.

Traditional Routes to Formic Acid

Aquilo and Horlenko II] of Celanese Chemical Co., Inc.,

Corpus Christi, Texas and Czaikowski and Bayne I2] of Scientific
Design have recently discussed in great detail the traditional
methods of manufacturing formic acid. Historically, formic acid
has been manufactured by (a) the liquid phase oxidation of hydro-


carbons to produce acetic acid, with formic acid being recovered

as a by-product, (b) the reaction of sodium formate with an inor-
ganic acid, and (c) the acidolysis of formamide with sulfuric acid.

The bulk of the U.S. formic acid production in recent years

has been via by-product from the liquid phase oxidation of butane
to acetic acid. With the advent of other routes to acetic acid
that are based on cheaper raw materials and hence more economic
than hydrocarbon oxidation, it is not likely that expansion will
occur in acetic acid, and hence formic acid, capacity by this
route. Currently, the leading route to acetic acid is Monsanto's
low pressure carbonylation of methanol in which no coproducts are

Future formic acid expansion in hydrocarbon oxidation plants

by the addition of purification facilities for its recovery might
be installed to achieve somewhat higher productivity when reaction
conditions are selected for maximum formic acid production. This
expansion is probably marginal at best. Clearly, the importance
of by-product formic acid from oxidation of hydrocarbons will
decline in the coming years.

The sodium formate route to formic acid depends on the salt

being available as either the by-product of pentaerythritol manu-

4HCHO + CH3 CHO + NaOH -+ C(CH20H)4 + HCOONa (1)

(Formal- (Acetal- (Sodium (Penta- (Sodium
dehyde) dehyde) Hydroxide) erythritol) Formate)

or as a product of the reaction between sodium hydroxide and

carbon monoxide under pressure.

CO + NaOH -+ HCOONa (2)

(Carbon (Sodium (Sodium
Monoxide) Hydroxide) Formate)

Adding a strong acid, such as sulfuric acid, to the sodium formate

liberates the formic acid and produces essentially stoichiometric
quantities of sodium sulfate as byproduct (equation 3). Companies

2 HCOONa + H2 S04 2 HCOOH + Na 2S04 (3)

(Sodium (Sulfuric (Formic (Sooium
Formate) Acid) Acid) Sulfate)

having access to by-product salt may continue to use this process,

but in the absence of an inexpensive source of sodium formate,
this process is not competitive beacuse it requires the consump-
tion of CO, caustic and sulfuric acid, with the need to dispose of
byproduct sodium sulfate.

In the production of formic acid via formamide, the formamdde

may be produced by the reaction of aunnonia and carbon monoxide at
high pressure in an alcoholic medium utilizing an alkaline cata-

NH3 + CO -+ HCONH (4)

(Ammonia) (Carbon (Formamide)

The formamide intermediate then reacts with dilute sulfuric acid

to yield formic acid and aunnonium sulf~te.

2 HCONH + H2 S04 + 2H20 -+ 2 HCOOH + (NH4)2S04 (5)

(Formami3e) (Sulfuric (Water) (Formic (Ammonium
Acid) Acid) Sulfate)

This process has been practiced on an industrial scale in

Europe, where formamide is 'also used to manufacture HCN. The
future of this route in Europe will, therefore, be affected by
the hydrogen cyanide demand and a significant factor will be the
demand for HCN in the production of acrylonitrile. Generally,
the high price of formamide as obtained from the small scale
plants currently available and the need of an outlet for byproduct
sulfate, makes this route anything but attractive. Also, the
recently reported teratogenicity of formamide [3] calls for strin-
gent environmental and safety precautions.

Conventional Uses of Formic Acid

The major uses for formic acid are in the textile
and leather industries. In Europe, formic acid has been in use
as a silage preservative for many years. The prospect for greatly
increased worldwide formic acid demand for agricultural purposes
has been raised in recent years.

Formic acid would be employed as a direct acidification agent

for ensiled forages. Silage is defined as animal feed resulting
from either a) the traditional anaerobic preservation of green
forage or other foodstuffs by limited fermentation of the ensiled
mass to produce lactic acid or b) the direct acidification of
green forage.

Acidification permits direct ensilement of wet fodders, thus

obviating dependence on weather for drying or the expense of
forced drying and resultant losses. The application of acids
decreases the degradation of ensiled material and allows a higher
animal intake, both factors resulting in higher animal production
(i.e. gain in weight) and milk production per ton of ensiled
forage. Preservatives, therefore, offer a direct increment to

the world's food supply. Those regions of the world having short
growing seasons and hence long periods when livestock are fed
stored forage would exhibit the greatest desire for direct acidi-
fication. Typical areas which would become markets for formic
acid for these purposes are Canada, parts of the u.s. and Northern

Formic acid is not alone as a viable silage preservative.

Others in the group are sulfuric acid, propionic acid, and mix-
tures of formic acid with formaldehyde. Sulfuric acid, which is
much less expensive than formic acid per hydrogen ion equivalent.
has satisfactorily substituted for formic acid in mixtures with
formaldehyde, and additional studies are being conducted. Earlier
studies indicated that sulfuric acid by itself leads to signifi-
cantly reduced forage consumption and can cause acidosis in live-
stock if excessive quantities are fed. A significant difference
between these moieties is that sulfuric acid does not volatilize,
while formic acid does. This means that by using precaution while
applying formic acid, it is possible to avoid spills and seepages
of the stronger sulfuric acid from silos. Such incidents have
caused problems that would not be encountered in the use of formic

Propionic acid, although an excellent preservative, is not

likely to find wide acceptance for hay crop preservation because
of its high cost. Propionic acid has been found to perform signi-
ficantly better than other preservatives in the prevention of mold
growth when used to treat corn and grains. As these crops exhibit
a higher value per pound, propionic acid has been accepted as the
agent of choice despite its price.

Formaldehyde is less costly than formic acid and performs

better as a retardant of protein degradation; however, when used
by itself, it performs poorly in preserving carbohydrate values
and decreases fodder digestibility slightly. Mixtures of the two
chemicals have been tested with su~cess, but it is still uncertain
whether the less costly mixture will yield results as good as
those obtained with formic acid alone.

A further complication with widespread formaldehyde usage,

unlike formic acid, lies in the fact that it must be precisely
measured as applied in order to ensure that protein digestibility
will not be adversely affected. A more serious consideration
which must be accounted for before wide use is made is the suspec-
ted carcinogenicity that has recently brought intense scrutiny by
government regulatory agencies.


With prospects being favorable for growth of future formic

acid demand, and with the growth of traditional formic acid manu-
facturing routes apparently unable to keep pace, attention was
focused on the development of a formic acid process that would
use cheap raw materials in a plant of modest capital investment.
Scientific Design and Bethlehem Steel, working together, have
developed a process that achieves formic acid synthesis with a
net consumption of carbon monoxide and water, and has minor
requirements for makeup methanol and catalyst. In this process,
methanol is carbonylated to methyl formate at moderate to high
pressure, and subsequently the ester is autocatalytically hydro-
lyzed to produce formic acid and methanol, the latter being recy-
cled to the carbonylation step.

1,1}dle working together on another process in which methyl

formate was produced as an intermediate from CO containing gases,
SD/Bethlehem carried out a program that completely defined the
methyl formate synthesis. Experimental work led to a complete
definition of the effects of major process variables on reaction
equilibrium and kinetics. These variables included temperature,
CO partial pressure, agitation level, and methanol, methyl formate
and catalyst concentration. Quite naturally, interest then grew
in the possibility of effecting the development of a spin off
process - the hydrolysis of methyl formate to formic acid.

Initially, based on our experience with ester hydrolysis

processes, it was anticipated that an acid catalyst would be
necessary in order to achieve economic hydrolysis rates. However,
it was discovered that the hydrolysis reaction is autocatalyzed by
the formic acid product, and even though the initial rate is quite
low, the rate increases significantly as the formic acid concen-
tration reaches catalytic levels. The use of this autocatalytic
effect of the formic acid product of hydrolysis has been patented
Further research and development efforts were expended to
study the potential for re-esterification of formic acid to methyl
formate and the decomposition of formic acid in the distillation
train. Optimal operating conditions were established that mini-
mize these reactions and allowed for maximum heat economies in
the flowsheet. Materials selection followed a thorough testing
of corrosion coupons in the reattion and purification environments.
These corrosion studies were conducted both batchwise and in
continuously operating equipment.

Synthesis Gas Source

There are a variety of sources of synthesis gas, but the two

main methods of production are steam reforming and partial oxida-
tion of gaseous or liquid hydrocarbons, or even coal. The CO
content of the synthesis gas product depends upon the type of
hydrocarbon in the feed. Partial oxidation of residual fuels can
result in CO concentrations as high as 50 volume percent. The
primary chemical consumers of synthesis gas are in the production
of methanol and of ammonia, and the amount of synthesis gas used
in these plants is extremely large relative to the CO requirements
of the formic acid plants that are likely to be built to satisfy
the growing agricultural market. It is conceivable that potential
formic acid producers in some locales will be able to obtain
incremental CO from existing methanol and ammonia facilities or
have synthesis gas assigned in a future methanol project. In
small or developing nations, such as Iceland or India, this will
not be the case, and potential producers in these areas will be
prime candidates for small dedicated CO generators. The off-gas
of carbon black plants or a blast furnace is low in carbon monox-
ide, but could also be available as a byproduct from existing
facilities that would ordinarily flare it.

Usually, impurities (but not n~cessarily inerts) in crude

synthesis gas must be reduced to satisfy downstream process
requirements. The SD/Bethlehem Steel formic acid process can
economically use synthesis gas streams that contain as little as
50 percent CO, as long as contaminants such as H20, CO 2 , O2 and
s·.tllfur compounds have been reduced to ppm levels. Inerts such as
H2 and N2 present no problems to the process. There are several
processes for upgrading the CO concentration in a synthesis gas
stream by either removing impurities or selectively recovering CO.
These include the Pressure Swing Absorption (PSA) of Union Carbide,
the hollow fiber permeation system (PRISM) of Monsanto and the
COSORB process offered by Tenneco. In a particular situation,
the quality of the available synthesis gas and the economic
alternatives will dictate which preconcentration process, if any,
is preferable.

Process Description

Figure I is a flowsheet of the Scientific Design/Bethlehem

Steel formic acid process. Note that the small amounts of makeup
catalyst and methanol are not shown.

Our process consists of a methyl formate synthesis section in

which methanol is carbonylated in the presence of a catalyst to
yield methyl formate,



Figure 1. Processing route for converting carbon monoxide into

formic acid.

CH OH + co + HCOOCH 3 (6)
(Met~anol) (Carbon (Methyl
Monoxide) Formate)

followed by an hydrolysis section where the ester is hydrolyzed to

yield formic acid and recyclable methanol. Product formic acid is


(Methyl (Formic (Met~anol)
Formate) Acid)

in subsequent distillations. Since methanol is recovered for

internal recycle the overall process chemistry is:

CO + + HCOOH (8)
(Carbon (Formic
Monoxide) Acid)

Methyl Formate Synthesis

Synthesis gas of the selected quality, is compressed from the

battery limits delivery pressure for feed to the reactor system.
In a typical plant where the available synthesis gas contains more
than 90 mol percent CO, .a single backmixed reactor is used. For
CO concentrations in the synthesis gas feed below this, and as low
as 50 mol percent CO, a two stage countercurrent backmixed reactor
system is preferred. For simplicity the f10wsheet of Figure 1
presents only the former situation.

In the reactor system, the CO contacts recycle methanol in an

agitated environment and is converted to methyl formate. The gas
leaving the reactor contains. most of the inerts that entered with
the feed gas as well as unreacted CO, and it is passed through a
vent condenser where methanol and methyl formate vapors are con-
densed and recovered. The high pressure vent gas is then sent
outside battery limits. Reactor temperature control may be accom-
plished in many ways, but pump around cooling appears to be the
most economically attractive alternative.

Net reactor effluent, which contains methanol, methyl formate,

catalyst, catalyst degradation products, dissolved inerts and CO,
is fed to a methyl formate-methanol separation system. Recycle
methanol and methyl formate from the hydrolysis product separation
are also fed to this column. Methanol containing active and spent
catalyst is taken from the bottom of the tower and recycled back
to the methyl formate synthesis reactor. It is from these tower
bottoms that any required spent catalyst purge should be taken
when necessary. With typical feed gases, less than a 1% b10wdown
of this stream is necessary. Make-up methanol and catalyst would
then be introduced into that portion of these tower bottoms, which
are being recycled.

The overheads from the methanol-methyl formate column, contain-

ing mostly methyl formate with some methanol, are sent forward to
the hydrolysis reaction section. Non-condensib1es that enter this
tower system are vented after suitable vent condensation has been
applied to recover the optimum methyl formate and methanol vapors
that would be present after the usual condensation has been accom-
plished. The tower pressure has been optimized taking into consi-
deration all of the salient factors involved in this system.


In the hydrolysis reactor, water and methyl formate react

autocatalytica1ly to give methanol and formic acid. The reactor
itself is a vertical internally baffled vessel that approaches
plug flow operation.

The reactor effluent is fed to a tower that separates it into

an overhead methanol/methyl formate stream and a bottoms fraction
containing water and formic acid. The overheads are returned to
the front sections via the methyl formate-methanol separation
column. This tower has been optimized with regard to losses via
re-esterification and thermal decomposition.

The water-formic acid stream is fed to a product purification

tower where the water recycle is taken overhead and returned to
the hydrolysis unit. Bottoms from the purification unit consist
of specification quality formic acid (85 wt percent formic solu-
tion). This tower design has been optimized with regard to formic
acid decomposition losses and the overheads are condensed by
reboiling other columns on the flowsheet and in preheating the
hydrolysis reactor feed.


As noted previously. carbon monoxide can be generated from a

number of feedstocks. Depending upon local circumstances, the
selection of the CO feedstock can protect the formic acid producer
using this technology from the vagaries of world hydrocarbon
pricing. There is little doubt that light hydrocarbons and naphtha,
presently the source of most formic acid production, will continue
to rise in price. The distinct advantage of the Scientific Design/
Bethlehem Steel CO based process is that it is less susceptible to
broad hydrocarbon price increases.

This process also has the advantage of becoming part of a

carbon monoxide consuming group of plants, e.g. methanol, acetic
acid, TDI, acetic anhydride, vinyl acetate, etc., with accompany-
ing economies of scale. An extension of this concept is to place
the formic acid plant "piggy back" on a large methanol or ammonia
plant. The advantage of this is that these plants are able to
take the high pressure vent gas.

Elements of production and capital cost estimates are shown in

Table 1 for plants that take either a 97 mol percent CO feed gas
or a 50 mol percent CO feed gas. For either situation, ppm levels
of active impurities are present. You can see that the capital
penalty associated with use of the leaner synthesis gas feed is
less than about 10 percent of the overall project BLCC. Catalyst
and chemicals costs are about 25% higher for the lean gas case
reflecting the higher level of active impurities per unit of CO.
converted. Cooling water usage is only minimally affected by the
difference in feedstock material, also due to higher flows.

Table 1. Scientific Design/Bethlehem Steel Formic Acid Process

Preliminary Economic Data

Basis: Plant Capacity - 20,000 metric tons/yr (contained in 85

wt % solution)
Operating - 8,000 hr/yr
Location - U.S. Gulf coast
CO Content of Feed Gas 97 mol% 50 mol%
Battery limits capital cost
$10,900,000 $11,900,000
(1st Quarter 1983)
Raw material consumption
per metric ton formic acid
Feed gas, NCuM 660 1530
High Pressure vent gas, NCuM 67 945
CO concentration of high pressure
75% 20%
Catalyst and chemicals per
8.05 9.95
metric ton formic acid, U.S.$
Utilities per metric ton
formic acid
Steam (18 kg/cm 2g), Metric Ton 4.5 4.5
Steam (7 kg/cm 2g), Metric Ton 2.5 2.5
Cooling Water (8T + 6°C) CuM 600 617
Power, kwh 155 303
Process water, CuM 0.6 0.6
Condensate'return (credit), CuM (6.3) (6.3)

Operators per shift 2 2
Foremen per shift 1 1
Supervisor 1 1

% of BLCC 4 4

The major increase in utilities associated with use of the

leaner synthesis gas feed is the power consumption, which is
approximately doubled. The power usage increase is due both to
the higher feed flow through the system and the fact that the
carbonylation step is run at a higher operating pressure with the
lean feed. Reactor staging and the higher operating pressure for
the lean gas case are the major factors contributing to the capi-
tal increase.

Our own evaluations indicate that the difference in transfer

price between these two situations is less than 3.0¢/kg of formic
acid product, assuming that feed and vent gases are valued equally
per mol in each case. This is a rather small margin to be able to
justify increasing the CO content from 50 to 97%, if the leaner
material is readily available.


We have shown that there is clearly a need for a new direct

route to formic acid. We have shown that, as we believe, the
growth in demand will be in the area of agricultural usage of this
material. We have shown that the market place for such new chem-
istry will be Canada, some parts of the U.S. and Northern Europe.
The newly developed Scientific Design/Bethlehem Steel formic acid
process, which takes the carbonylation route, is such a process.
It has been developed to handle a great variability in synthesis
gas feedstocks. This offers licensees of this technology a means
to escape from the ever escalating costs of hydrocarbon based feed-
stocks by judicious choice of the synthesis gas source.

The SD/Bethlehem process consumes essentially only CO and wat-

er, cheap raw materials. We have demonstrated that the capital
investment for such a plant, regardless of feedstock source, is
rather modest. We have demonstrated that every effort has been
made to optimize the operating conditions, and that maximum heat
and material economies have been incorporated into our design.
An added feature of our process is that no foreign substances,
chelating agents or solvents, are employed, and therefore there
is no danger that agricultural use will lead to harmful effects
either to the environment or to living things. This is a signifi-
cant factor, especially since it is well known that animals are
very sensitive to the presence of trace quantities of these types
of materials and the concomittant change in their feeding habits
could be disastrous.

We look forward to the early commercialization of this

technology. Presently, several projects are in the negotiation


1. A. Aqui10 and T. Hor1enko, Hydrocarbon Process., 59(11):120,

2. M. P. Czaikowski and A. R. Bayne, Hydrocarbon Process., 59(11):
103, (1980).
3. Material Safety Data Sheet, BASF Wyandotte Corp., Wyandotte,
Michigan 48192.
4. J. B. Lynn, O. A. Homberg, and A. H. Singleton, U.S. Patent
3,907,884 (1975); assigned to Bethlehem Steel Corp.

w. R. Pretzer and M. M. Habib

Gulf Research & Development Company

P.O. Drawer 2038
Pittsburgh, PA 15230


Alcohol homologation is a general term describing a process in

which an alcohol with n carbon atoms is converted into an n+l carbon
alcohol, i.e. an additional methylene unit has been added to the
starting alcohol. As commonly used, alcohol homologation refers to
the reaction of an alcohol with carbon monoxide and hydrogen to give
higher carbon number oxygenates:

ROH + CO + 2H2 ~ RCH20H + H20.

The reaction was reported by several groups during the 1940's

[1-3]. During most of the following three decades, alcohol homolo-
gation received little attention from industrial and university re-
search groups. In the mid-1970's industrial interest was revived as
efforts to develop non-petroleum routes to "petrochemicals" stimu-
lated a renaissance in synthesis gas chemistry. The reader is re-
ferred to several excellent reviews for detailed summaries and dis-
cussion pertaining to prior published results dating from the first
report of the reaction until the late 1970's [4-6].

Development of Alcohol Homologation Catalysts

The largest group of homogeneous alcohol homologation catalysts

contain cobalt as an essential component. Besides cobalt, the cata-
lyst systems can also contain promoters, modifiers, co-catalysts and
other additives. The specific identity and combination of these
other catalyst components, as well as their relative amounts are the
bases for most of the recent advances in catalyst development.


Prior to 1979, the number of reported cobalt homologation cata-

lysts and processes was limited. The early catalysts described [3,
7-9] consisted simply of cobalt carbonyl (or a precursor to the car-
bonyl under reaction conditions). These catalysts suffered from low
activity, poor selectivity, and questionable stability for a commer-
cially viable catalyst.

A major advance in reaction rate was achieved when iodine and

iodide promoters were added to the catalyst system in 1956 [10]. By
1966, the problem of selectivity was addressed through the use of
phosphorous modifiers [11] and ruthenium co-catalysts [12] in con-
junction with cobalt and iodide. These latter improvements were di-
rected towards methanol homologation and claimed enhanced selectivity
to ethanol.

Synthesis of predominantly acetaldehyde and acetals via methanol

homologation was reported in 1967 [13]. The catalyst was simply a
soluble cobalt compound and an iodine promoter. The fundamental
teaching of this patent was that yields of acetaldehyde and acetal
increased by maintaining a constant iodine content while lowering
the cobalt concentration. In other words, higher IICo ratios favor
acetaldehyde synthesis over ethanol production.

In 1968, an improved process for the homologation of normal al-

cohols was reported [14]. The catalyst was again simply a soluble
cobalt compound and iodine. However, addition of 0.5 to 1.0 mole of
water per mole of n-alcohol feed enhanced the rate of higher alcohol
homologation. Also, to enhance alcohol production, the IICo ratio
was kept low, at about 0.5.

The status of alcohol homologation processes and catalyst devel-

opment prior to the sudden growth in the late nineteen seventies was
sufficient to suggest many tantalizing catalyst possibilities and a
very rich reaction chemistry. Iodine and iodides not only promoted
the reaction, but apparently could also influence product selectiv-
ity. Product selectivity could also be altered by addition of phos-
phorous compounds to the basic cobalt carbonyl catalyst. While other
metals were not as effective as cobalt for catalyzing the homologa-
tion of alcohols, their presence could alter the final product dis-
tribution. Solvent effects might be strong. Water, a co-product of
alcohol homologation, was shown to promote the reactivity of higher
alcohols when mixed with the feed alcohol.

This report describes a study of just one of the many factors

influencing alcohol homologation, i.e. iodine and iodide promoters.
The catalyst systems studied were restricted to iodine and iodide
promoted cobalt and cobalt-phosphine compositions. Much of the prior
research on alcohol homologation has used methanol exclusively as the
feedstock. This has been motivated by the attractiveness of metha-
nol as a readily available, low-cost, synthesis gas derived interme-

diate. In this study, we have evaluated a series of alcohols of

varying steric and electronic properties in order to gain insight
into the manner in which iodine and iodides mediate alcohol homolo-

A Caveat

Alcohol homologation is a complex reaction producing a range of

products with widely varying chemical and physical properties. Meth-
anol has been the most extensively studied. Accordingly, it is ap-
propriate to illustrate the complexity of the homologation process
by detailing the products observed when methanol is reacted with
synthesis gas.

The direct hydrocarbonylation products of methanol are ethanol

and acetaldehyde. Typically, 50-90% of the methanol reacted will
form these products. Further homologation of product ethanol gives
propanol and propanal at 2-15% selectivity. Aldol products arising
from subsequent condensation of the primary product acetaldehyde in-
clude n-butanal, 2-ethylbutanal, 2-ethylhexanal, alcohols correspond-
ing to these aldehydes, and unsaturated aldehydes derived from the
dehydration of the hydroxy-aldehyde aldol intermediates. These prod-
ucts can be present in amounts as little as 0.1% to quantities rang-
ing up to 40%.

Carboxylic acids and their esters are a major group of by-prod-

ucts. Acetic acid and acetate esters can arise by direct methanol
carbonylation, while other esters can be formed by a series of ester-
ification reactions (e.g. direct esterification, inter-esterifica-
tion, and transesterification). Formates can also be produced by
direct carbonylation processes. While a score of such products can
be formed, the only major ones observed are methylacetate (3-30%),
ethylacetate (2-10%) and methylformate (0.1-2%).

Alcohol dehydration gives ethers, predominantly dimethylether

(2-35%), diethylether (1-15%), and methylethylether (1-5%). Hydro-
genation reactions produce hydrocarbons corresponding to the alcohol
species present. Thus, methane is the predominant hydrocarbon ob-
served and can represent 5-15% of the methanol reacted. When ethanol
is the major homologation product, ethane is also seen in the range
of 1-3%.

In summary, the homologation of the simplest alcohol, methanol,

can yield higher alcohols, aldehydes, carboxylic acids, esters,
ethers, and hydrocarbons. Typically, four or five compounds comprise
>90% of the products.

Experimental procedures, particularly during product recovery

and work-up, must be designed to quantitatively determine all prod-
uct frat ions and obtain a precise mass balance. This is most impor-
tant with respect to accountability of methanol charged. Depressur-
ization of the reactor is a crucial step during which volatiles such
as methane, ethane, carbon dioxide, and dimethylether can easily be
lost. Both the temperature and rate' of venting are important. If
the venting is too rapid, liquid products are entrained and lost. If
the temperature is too high, light products are selectively distilled
from the reaction mixture, skewing the distribution of liquid prod-
ucts to the higher boiling components. Simple analysis of the off-
gases is inadequate unless the volume of vented gas is also deter-
mined. Further, when the off-gas contains significant quantities of
less volatile compounds, selective condensation of these materials
may occur in process lines prior to analysis resulting in inaccurate
product selectivities.

Care must be exercised if an internal GC standard is used. The

standard generally remains with the liquid product, and is of no use
in analyzing the off-gases. The homologation of methanol produces
water, which can result in phase separation if the standard is a non-
polar substance. The unequal distribution of products and standard
between the phases significantly reduces the utility of the tech-

In comparing literature results, attention must be given to the

experimental technique and analytical methods employed. For example,
catalysts comprising high iodine to cobalt loadings, often produce
significant amounts of dimethyl ether and may give a product that is
40% dimethylether and 50% ethanol. If the reactor is simply vented
at 25°C and no provisions made for accurate off-gas analysis, the
product (sans ether) will appear to be 83% ethanol - a selectivity
dramatically different from reality.


Materials and Procedure

All alcohols used in this study were reagent grade, purchased

from Baker or Aldrich. Cobalt carbonyl, cobalt(II) iodide, iodine,
and sodium iodide were purchased from Alpha. The cobalt carbonyl was
stored in an inert atmosphere dry box at -32°C at all times. Carbon
monoxide/hydrogen gas mixtures were supplied by Airco. All chemicals
were used as received without any further purification.

All experiments were performed in a 300 cc, 316 SS Autoclave

Engineers vessel rated for 5100 psig and 650°F. Synthesis gas was
pressured into a reservoir by a Haskell air-driven compressor. The
gas flow from the reservoir to the autoclave was controlled by a
Moore 55 nullmatic controller, a Bristol pressure transmitter, and
an Annin valve fitted with AA trim. This assembly permitted gas flow
into the autoclave upon demand at constant pressure. The gas uptake

from the reservoir was monitored by a Sensotek pressure transducer

with a digital read-out.

In a typical experiment, the desired amounts of alcohol and

catalyst were weighed and charged into the autoclave; the unit was
then closed, sealed, pressure tested for leaks with nitrogen and then
twice purged with nitrogen followed by synthesis gas. The unit was
then brought to 2000 psig with synthesis gas. The autoclave was
heated with an (ECS) controlled heating mantle. When the reaction
temperature was reached, (usually 10-15 min), the pressure was ad-
justed to the desired value and maintained at this level throughout
the entire reaction period. At the end of this period, heat and gas
flow were terminated and the contents of the autoclave were cooled to
about -70°C with an external acetone/dry ice bath. The reactor was
slowly vented to atmospheric pressure through a dry test meter to·
measure the volume of gas released. All of the vented gas was col-
lected in a 70 liter multilayered sampling bag and then sampled for
GC analysis. The recovered liquid product was weighed and then ana-
lyzed by gas chromatography.

Product Analysis

Liquid product was analyzed with a Hewlett-Packard Model 5730A

gas chromatograph using a thermal conductivity detector and 16' x
1/8" Porapak Rand 8' x 1/8" Porapak Q SS columns connected in se-
quence. A temperature program for analyzing the oxygenates was used
with an initial holding time of 20 minutes at 35°C, heating at a rate
of 15 degrees/minute to 60°C, then at a rate of 6 degrees/minute to
a final temperature of 220°C. For the higher alcohols, a 12' x 1/8"
SS column packed with 10% OV-lOl on 80/100 mesh Chromo sorb W, was
also used.

The gas analysis was performed using a Carle series 500 CGC with
dual detectors. The oxygenates were analyzed by a flame detector
while H2' CO, C02' 02, N2, and light hydrocarbons were analyzed by a
thermal conductivity detector. The whole analysis was performed by
either a single manual or in-line injection using a set of seven dif-
ferent columns. Valve switching as well as temperature programming
were controlled by a microprocessor.

Rate Data Procedure

Data for initial gas uptake studies for different catalyst sys-
tems were collected in the unit described above. However, for better
accuracy in measuring the gas uptake, a small reservoir with a volume
of 120 cc was used. Sensotek pressure transducers fitted with dig-
ital read-outs were connected to both the reactor and the reservoir.
The temperature of the reservoir was held constant during an exper-
iment, while the temperature of the autoclave varied ±1.5°C from the
desired reaction temperature. Time and gas uptake readings were

recorded from the moment the unit reached desired reaction condi-


Methanol Homologation

Iodine and compounds containing iodine have been recognized

since 1956 [10] as promoters for alcohol homologation using cobalt
catalysts. These promoters can be generally classified as either
strong or mild activators. The former comprise such substances as
12' CH3I, and HI, and generally possess considerable covalent char-
acter. Under reaction conditions, they are capable of forming HI.
Representative members of the latter group include NaI, LiI, KI,
CaI2, and SrI2 and are characterized as being ionic.

In the homologation reaction, a strong activator like iodine can

act in several ways. Its primary function is to promote the rate of
alcohol conversion. Product selectivity can also be altered by vary-
ing the amount of promoter either by changes in specific identity of
the soluble cobalt catalyst or by changes in the acidity of the re-
action medium. In the latter case, the activator is a source of HI
and can promote alcohol dehydration and acid catalyzed aldol conden-

A study of the effect of iodine and iodide promoters was per-

formed using the homologation of methanol as the test reaction. Sev-
eral catalyst systems were used and the loading of iodine promoter
relative to cobalt was varied. The rate of synthesis gas consumption
and the rates of formation of individual products were determined.

Promotional Effects

In Table 1 are shown the results of a series of experiments in

which the IICo ratio was varied from 0.0 to 10.0. The catalyst was
a ligand-free system composed of cobalt carbonyl and elemental
iodine. The reaction time was held to 20 min to minimize secondary
reactions. The cobalt concentration was 0.1 M and the reaction con-
ditions were 4000 psi and 200°C.

The promotional effect of a strong activator, like iodine, is

readily apparent by considering the rate of product formation and
synthesis gas consumption as the IICo ratio is progressively changed
from 0 to 10 (Table 1 and Figure 1). With no iodine in the system,
the homologation of methanol proceeds slowly and after a reaction
time of 20 min only 12% of the methanol is converted. The turnover
frequency of the catalyst (gram moles of carbonylated product per
gram atom of cobalt per hour, where carbonylated product is the sum
of the moles of acetaldehyde, ethanol, methylacetate, and their de-
Table 1. Effect of Iodine/Cobalt Ratio on Product Selectivity over a Cobalt Carbonyl Catalyst a z

I/Co mmol products based on methanol mmol :x>
Methanol % HAc & Deriv. c r
Molar ()
Conversion (based on HAc) o
Ratio Me20 HAc EtOH MeOAc CH4 Others b
0.0 12 58 61 20 32 49 4 61 o
0.5 36 69 125 105 94 59 417 221 s::o
1.0 47 77 216 108 142 68 464 284 r
110 200 528 713
2.0 69 523 6 75
4.5 64 305 375 2 177 81 408 663 :x>
10.0 68 834 130 0 124 121 466 314 6

aCatalyst = Cobalt carbonyl = 5 mmol; 12; Methanol 100 cc

Pressure = 4000 psig, H2/CO = 1
Temperature = 200°C
Time = 20 minutes
bMethyl formate, dimethoxy ethane, propanal, butanal, and higher aldols
cHAc + dimethoxy ethane + n(aldols) where n = HAc units

4600 0 1- 0
~c~o~ .
4000 0 It..c°- t ...-'1
'V ~ClE.2_ •
8BOO 0 I!..CO=4.5 __ 'V""
0 It..Co=tO
~ .000 --
01( 2600

~ 2000




0 2 4 II
• 10 12
14 18 11 20

Figure 1. Initial rates of synthesis gas uptake for methanol/cobalt/


rivatives) is only 34. Adding iodine to a level of I/Co = 0.5 re-

sults in an almost four-fold increase in the rate of homologation and
a catalyst turnover frequency of 126. Further additions of iodine
to I/Co ratios of 1.0 and 2.0 give still larger turnover numbers of
175 and 276, respectively. At this latter value, homologation of
methanol is proceeding about eight times faster compared to the un-
promoted catalyst.

The rate of synthesis gas consumption was also measured as a

function of the I/Co ratio using the same cobalt carbonyl and iodine
catalyst system. The results are plotted in Figure 1. The unit of
synthesis gas consumption is given in pounds per square inch (psi)
and is the measured drop in pressure in a high pressure reservoir
feeding synthesis gas on demand to the reactor vessel. Accordingly,
gas consumption is relative and the absolute amount of gas consumed
was not measured.

As might be expected, the pattern of total gas consumption as a

function of I/Co ratio parallels the trend seen for actual product
formation. In both cases the maximum occurs when I/Co = 2. Addition
of still higher amounts of iodine to give I/Co = 4.5 increases the
initial rate of synthesis gas consumption by a factor of eight rel-
ative to the iodine-free system. An apparent saturation in the pro-

motional effect of a strong activator is reached at this level as a

further increase in iodine leading to IICo = 10 does not increase the
initial rate of synthesis gas consumption.

Total synthesis gas consumption and, accordingly, the absolute

amount of homologated product are less at these high IICo ratios than
for IICo = 2. The quantity of dimethylether produced when IICo = 4.5
and 10 is very high and contributes significantly to the total pres-
sure on the reactor. Thus, while the total pressure is held at 4000
psi during the batch reaction, the partial pressures of carbon mon-
oxide and hydrogen are considerably below that value during the
latter stages of the experiment resulting in reduced reaction rate.
The high temperature and hydrogen partial pressure encountered under
reaction conditions are conducive to the conversion of some of the
strong promoter to hydrogen iodide with a resultant increase in the
acidity of the reaction medium. As a consequence, acid catalyzed
dehydration of methanol becomes a competitive reaction pathway.

Similar observations regarding the promotional effect of iodine

can be made using a catalyst to which a chelating phosphine, i.e.
bis(1,2-diphenylphosphino)ethane (dppe) has been added (Table 2 and
Figure 2). When IICo = 0.5, 2.0, and 4.0, the catalyst turnover fre-
quencies are 96, 433, and 279, respectively. The optimal IICo is 2,
which is the same as seen for the phosphine-free system. A catalyst
composed of cobalt(ll) acetate, HI, and triphenylphosphine was also
reported to give the best yield of homologated product at IICo = 2
[15]. The initial rate of synthesis gas consumption also shows a
maximum at IICo = 2. On either side of this ratio, lower rates are
seen. This is unlike the ligand-free system in which the initial
rates continue to increase with increasing IICo until a saturation is
reached at IICo = 4.5, above which the initial rate is constant.

Selectivity Effects

The relative distribution of ethanol and acetaldehyde changes

as a function of the IICo ratio. As shown in Figure 3, almost equal
amounts of ethanol and acetaldehyde are produced at IICo = 0.5. In-
creasing the IICo ratio to 1.0 approximately doubles the amount of
acetaldehyde formed, while the quantity of ethanol produced remains
unchanged, i.e. the rate of aldehyde reduction to alcohol is con-
stant. When IICo = 2, almost four times as much acetaldehyde is
formed than when IICo = 0.5. However, there is now a precipitous
drop in the amount of ethanol formed and the ability of the catalyst
to reduce aldehyde is strongly suppressed. It is significant to note
that production of total homologated product is greater for IICo = 2
than for IICo = 1, even though ethanol becomes a minor product at the
higher IICo value. Iodine still shows a promotional effect at IICo =
2, although aldehyde hydrogenation to alcohol has been inhibited.
At higher IICo ratios, the catalyst is still unable to readily hydro-
genate acetaldehyde to ethanol.

Table 2. Effect of Iodine/Cobalt Ratio on Product Selectivity for the Catalyst Coba1t/lodinel

I/Co mmo1 products based on methanol mmo1

Methanol %
Molar HAc & Deriv. c
Ratio Me20 HAc EtOH MeOAc CH4 Others b (based on HAc)

0.5 23 41 113 21 88 64 284 205

2.0 81 88 897 20 300 69 377 1124
4.0 71 332 608 0 153 83 330 769

HCata1yst = Cobalt carbonyl = 5 mmo1; 12; dppe 5 mmo1; Methanol = 100 cc ?E

Pressure = 4000 psig, H2/CO = 1
Temperature = 200°C ::0
Time = 20 minutes m
bMethy1 formate, dimethoxy ethane, propana1, but anal , and higher a1do1s N
cHAc + dimethoxy ethane + n(a1dols) where n = HAc units


IlOO 0 I/Co-<J.5 :,,10.'11 Cobl1ll ~I
1~o.~C. 4000"$(11
[,. !fPo..=..2_ p/ Co. ' .H2/CO='

11000 0 !!t;.~.":.L ....


~ 1500
'iCI 1000




Figure 2. Initial rates of synthesis gas uptake for methanoll


A more subtle change is seen as IICo is raised from 2.0 to 10.

Simple carbonylation of methanol to methylacetate declines 38%, while
hydrocarbonylation to acetaldehyde decreases by 75%. Even consider-
ing the acetaldehyde that is converted to condensation products, the
total amount of acyl intermediate that is hydrogenated to aldehyde
compared to that which is esterified declines when IICo > 4.5. As
shown in Table 1, the ratio of hydrocarbonylated product to carbon-
ylated product is essentially constant at about 3.5 as IICo increases
from 0 to 4.5, but declines as IICo increases to 10.

Methane formation is also seen to be a function of the iodin~

loading, steadily increasing as IICo varies from 0 to 10 (Table 1).
No inflection or maximum at a particular IICo ratio occurs and it
appears that the increase in methane production is a function of ab-
solute iodine concentration. A simple HI catalyzed mechanism for
methanol hydrogenolysis to methane is possible [16] and the monotonic
increase in methane production as the iodine concentration increases
is consistent with this mechanism.

However, methane is formed even in the absence of iodine and

there is no abrupt increase in the rate of methane production when
iodine is introduced. Recall that the rate of homologation increased
by a factor of almost four (compared to the no iodine system) when


100 I!!!.-
w_ ........ .
«~U.Il~.... D "d(m."M:Z:~8t,\un. + 'ft cUdols
lOa n-HAc units
' .

u 000

..." ..,....•......, ..
£ eoo
E ......
saO ·····························.. 0



e ~
4 15
IICo Moler Retio
Figure 3. Effect of iodinelcobalt ratio on product selectivities

iodine was added to give IICo = 0.5. Methane formation increased

only 20% upon the addition of iodine. We are inclined to believe
that methane production is related to reaction of a CH3-Co complex
with either H2 or HI and may proceed through a Co(III) intermediate.
Reductive elimination of methane from a methyl metal hydride has been
shown to proceed extremely fast [17] and the hydrogeno1ysis of meth-
anol using iodine-promoted rhodium complexes has recently been re-
ported [18].

Previously, it had been reported that the rico ratio can dramat-
ically influence the yield (a function of conversion and selectivity)
of products obtained by homologating methanol [19]. In this earlier
work, two catalyst systems were studied. One system, producing eth-
anol, comprised Co(acac)2' tri-p-tolylphosphite, and iodine, showed
low ethanol yields when IICo > 0.5. The other system, giving acetal-
dehyde, was composed of Co(acac)2, Ph3As, and iodine. Significantly
lower acetaldehyde yields were seen when rico> 0.25. It is impor-
tant to note that the reaction time for these experiments was 3 hr
compared to 20 min for the work described above.

The major consequence of prolonged reaction time is to increase

subsequent reactions and by-products formation, particularly those
involving a reactive molecule like acetaldehyde. The predominant
product found after reaction for 20 min is always acetaldehyde. Two
competing reactions involving acetaldehyde can predominate with
longer reaction time, i.e. reduction to ethanol and condensation to

aldol products. These two reactions are mediated by strong activa-

tors like iodine and probably proceed by Hr derived from such pro-

The observation that the yield of ethanol declines when rico >
0.5 is consistent with the selectivity pattern as a function of rico.
Hydrogenation ability is suppressed when rico > 2 and, while signif-
icant amounts of ethanol are produced when rico = 1, more acetalde-
hyde than ethanol is formed. With an extended reaction time, this
acetaldehyde is subject to condensation, a process more favorable
under the more acidic conditions prevailing when rico = 1 than at
rico = 0.5. To produce high yields of ethanol in minimal reaction
time, a catalyst capable of rapidly hydrogenating acetaldehyde is
required. Since cobalt cannot do this under conditions most favor-
able for rapid aldehyde synthesis, a cocatalyst is required. The so-
lution, as amply represented in the patent literature [12,20a-f], has
been the use of ruthenium compounds. Catalyst systems containing
ruthenium cocatalysts produce optimal ethanol yields when rico = 2,
which is the ratio giving the maximum rate of acetaldehyde production
using catalysts containing only cobalt.

The maximum yield reported [19] for acetaldehyde production in

3 hr using cobalt modified with Ph3As occurred at rico = 0.25 and not
at rico = 2, as would be expected from this current study, because
of the prolonged reaction time leading to significant generation of
aldol products. The observation that acetaldehyde and not ethanol is
the major product even at a low rico ratio is an example of the abil-
ity of ligands, in this case Ph3As, to alter catalyst properties.

Homologation of Higher Alcohols

A comparative study of the homologation of several alcohols was

conducted using a catalyst comprising cobalt(rr) iodide, a mild ac-
tivator sodium iodide, and bis(1,2-diphenylphosphino)ethane (dppe).
The relative ratios of the catalyst components were held constant for
the entire series of experiments at Co:r:dppe = 1:8.5:0.5 and the
cobalt concentration was held at 0.01 M.

rn Table 3 are shown the results obtained for the homologation

of methanol, ethanol, and 2-propanol. Also shown are results when
5 vol % water was charged with the alcohol feed. The dominant homol-
ogation product is aldehyde, though some alcohol is also formed as
a result of the 2 hr reaction time. Simple alcohol dehydration proc-
esses giving olefins and symmetric ethers are major side reactions.

The order of reactivity for homologation of the three alcohols

in the absence of added water is methanol » ethanol » 2-propanol.
Methanol is about 12 times more reactive than ethanol and 157 times
more reactive than 2-propanol. The reactivity of ethanol increases
by 29% with the addition of water to the feed, while the rate of

Table 3. Homologation of Cl-C3 Alcohols,a mmol of Products

Alcohol Aldehydes Hydrocarbons Ethers Others b

MeOH Ethanal = 1145 136 461 232

Propanal = 3
Butanal = 106
EtOH Propanal = 101 31 152 27
2-PrOH Butanal = 3 49 13 11
propanal = 5
EtOHC + 5%, H20 Propanal = 130 38 176 31
2-PrOHc + 5% H20 Butanal = 21 51 37 17
propanal = 29
aCatalyst = CoI2 ~ 1 mmol; NaI 6.5 mmol; dppe = 0.5 mmol; Alcohol 100 cc -I
Pressure = 4000 psig, H2/CO = 1 m
Temperature = 200°C »Z
Time = 2 hr 0
bhigher alcohols and condensation products s:
c95 cc s:

conversion of 2-propano1 jumps by about 6 times. Interestingly, the

inclusion of water with the alcohol charge did not significantly de-
crease the relative amount of alcohol dehydrated to ether.

Since alcohols containing a 5 vol % water are more reactive,

these systems were evaluated in greater detail. Table 4 shows the
results obtained using various primary, secondary, and tertiary al-
cohols. The catalyst system was again CoI2, NaI, and dppe at the
same ratios as used previously. Aldehyde was the dominant homolo-
gation product and ether and hydrocarbon were the major by-products.
Reactivity for homologation decreased in the order ethanol > 1-prop-
ano1 - 1-butano1 > 2-propano1 - 2-butano1 » t-butano1.

The site of homologation is strongly influenced by the position

of the hydroxy group on the feed alcohol. Thus linear, primary al-
cohols are converted mainly to straight chain aldehydes and alcohols.
Linear secondary alcohols give more branched aldehyde and alcohol
products in which homologation occurs at the site of the hydroxy
group. Straight chain aldehydes and alcohols comprise 71% and 83%
respectively, of the products resulting from homologation of 1-prop-
ano1 and 1-butano1. However, the straight chain products comprise
only 44% and 50% of'the products formed by homologation of 2-propano1
and 2-butano1, respectively. For t-butano1, 75% of the homologated
product was that expected if homologation occurred at the hydroxyl
site. Thus, while some rearrangement is seen, homologation proceeds
mainly via an intermediate dictated by the structure and hydroxyl
position of the starting alcohol.

Prior studies of the homologation of higher alcohols employed

catalysts comprising cobalt carbonyl (or a precursor to cobalt car-
bonyl) [3,8], and cobalt carbonyl promoted with iodine [10,14]. The
various alcohols examined were found to exhibit reactivity in the
order tertiary> secondary> primary. Anomalous behavior was ob-
served for the nondehydratab1e primary alcohols, methanol and benzyl
alcohol. These substrates were generally more reactive than second-
ary alcohols. Thus, the order of reactivity of alcohols using simple
cobalt and cobalt-iodine catalysts is tertiary > methanol > benzyl >
secondary> primary.

In this prior work, the specific products reported to have been

obtained from homologation of various higher alcohols suggest that
an olefin, derived from alcohol dehydration, could be an intermedi-
ate. Thus, n-butano1 and 2-butano1 each give n-pentanol and 2-
methyl-1-butano1 in approximately a 2:1 ratio [8]. A similar distri-
bution is obtained from hydroformy1ation of linear butenes using un-
substituted cobalt carbonyl catalyst. A labelling study employing
l4C enriched methanol has also supported the concept of an olefin
intermediate [21].

These observations have led to the proposal that the homo10ga-

Table 4. Carbonylation of C2-C4 Alcohols with CoI2/NaI/DPPE Catalyst,a mmol of Products m

Aldehydes Carbonylation
Alcohol + 5% Water Alcohols Hydrocarbons Ethers b Others

EtOH Propanal = 186 l-Propanol = 31 56 220 25 3.9

l-PrOH Butanal = 80 l-Butanol = 24 C3's = 58 171 23 2.5
2-Methyl- 2-Methyl-l-
propanal = 36 propanol = 7
2-PrOH Butanal = 41 l-Butanol = 14 C3's 71 45 11 1.8
2-Methyl- 2-Methyl-l-
propanal 59 propanol = 12
n-BuOH Pentanal = 106 l-Pentanol = 29 C4's 131 21 30 1.2
2-Methyl- 2-Methyl-l-
butanal = 9 butanol = 18
sec-BuOH Pentanal = 31 l-Pentanol = 17 C4'S 214 13 14 0.4
2-Methyl- 2-Methyl-l- ::§!
but~nal = 32 butanol = 16 ?J
t-BuOH 2,2-Dimethyl 2,2-Dimethyl- C4's 583 7 5 0.03 :D
propanal = 8 propanol = 4 -I
3-Methyl- N
3-Methyl-l- m
butanal = 3 butanol = 1 l>

aCatalyst = CoI2 = 1 mmol; NaI 6.5 nnnol; dppe 0.5 mmol; Alcohols 95 cc s:
Pressure = 4000 psig, H2 /CO = 1 s:
Temperature = 200°C I
Time = 3 hr !:!:!

tion reaction proceeds via a carbonium ion [3], and that an alcohol
dehydration mechanism is operative [8], at least for those alcohols
which can be dehydrated to olefins. The observed reactivity pattern
(3°>2°>1°) is consistent with an El elimination involving a proton-
ated alcohol giving a carbonium ion. Subsequent reaction of this
carbonium ion could involve nucleophilic attack by iodide or Co(CO)4,
or elimination of a proton to give an olefin. Hydroformylation of
olefin leads to product aldehyde.

Our results (vide supra) differ from this prior work. Using a
catalyst comprising CoI2' NaI, and dppe, the present work has ob-
served the order of reactivity to follow methanol > primary > second-
ary > tertiary. Further, much less rearrangement was observed, i.e.
the position of the hydroxy group in the starting alcohol was pre-
dominantly the position of addition of the -CH2- unit. These results
are consistent with an SN2 mechanism, most likely involving the for-
mation of alkyl iodide from alcohol and iodide.

The difference in behavior of the higher alcohols towards homol-

ogation can be attributed to the differences in catalyst composition.
In those systems showing greater reactivity for tertiary alcohols,
the cobalt concentration is high, the iodine concentration is low,
and the I/Co ratio is low. Under the conditions of reaction, HCo-
(CO)4 will form and, if 12 has been charged, HI will also be gener-
ated. The reaction medium is quite acidic and protonation of alcohol
should be facile. Subsequent loss of water, carbonium ion formation,
and proton elimination affords olefin. The observed El elimination
reactivity pattern is consistent with this scenario. The olefin is
then subject to hydroformylation since an excellent oxo catalyst,
HCo(CO)4, is present. Generally, the distribution of product alco-
hols is consistent with that observed during olefin hydroformylation
using simple, unmodified cobalt catalysts.

When a small quantity of iodine is added to the system, approx-

imately a 4-fold increase in catalyst turnover frequency is seen
[3,10]. This comparison is only qualitative since reaction condi-
tions were different and calculations are based only on gas adsorp-
tion and not actual product obtained. The promotional effect for
higher alcohol homologation under conditions of low iodine (strong
activator) to cobalt ratio is attributed to the presence of HI and
its capability to readily catalyze alcohol dehydration. It is impor-
tant to note that the I/Co ratio employed was 0.50. At this ratio,
most of the cobalt should be unaffected by iodine and a relatively
high concentration of the hydroformylation catalyst HCo(CO)4 should
be maintained.

The catalyst system studied in this current research is charac-

terized by a lower cobalt concentration and a higher iodide to cobalt
ratio (I/Co = 8.5). A mild activator, NaI, was used and a phosphine,
specifically dppe, was also present. The low cobalt concentration,

high IICo ratio, and the presence of phosphine should insignificantly

minimize, if not eliminate, the presence of HCo{CO)4' Some phosphine
substituted cobalt carbonyl hydride may be present.

In the presence of the CoI2, NaI, and dppe catalyst system, al-
cohol should still be protonated and olefin formed as described
above. However, olefin will accumulate since no effective hydrofor-
mylation catalyst is present. HCo{CO)4 is not present in any signif-
icant amount and cobalt carbonyl modified with aromatic diphosphine
is a drastically less active hydroformylation catalyst [22]. This
pathway to homologated product is effectively blocked. A second
pathway, however, is available, i.e. nucleophilic attack by 1- on the
protonated alcohol to give alkyl iodide. The mechanism for this
process will be SN2 and thus favors primary alcohols. Oxidative ad-
dition of alkyl iodide to a cobalt compound will give an alkyl-cobalt
complex and subsequent alkyl migration (or carbon monoxide insertion)
affords an acyl complex. Hydrogenation andlor reductive elimination
will give product aldehyde. This second pathway is not viable for
primary alcohols in iodine-free or -deficient systems. In the ab-
sence of iodine no strong nucleophile is present, while in those sys-
tems with low IICo ratios, the concentration of free 1- is low. Fur-
ther, for those systems promoted with 12, the iodide is probably
present as HI, a considerably weaker nucleophile than 1-.

The contrasting properties of covalent and ionic iodine activa-

tors are clearly seen by comparing the results shown in Tables 4 and
5. The cobalt concentration, reaction time, pressure, and tempera-
ture were the same for all experiments shown in the two tables, and
(with one exception) the IICo ratio was held at 8.5. With a strong
activator, 12, and IICo = 0.5, t-butanol is readily converted to the
rearranged product, 3-methylbutanal (Table 5). Only about 10% of the
homologated product is 2,2-dimethylpropanal. Raising the level of
12 to IICo = 8.5, drastically reduces the amount of t-butanol homol-
ogated by about 93%. However, rearranged aldehyde and alcohol still
comprise about 90% of the homologated products. Little 5-butanol
is homologated using a mild activator NaI (I/Co = 8.5); however, the
aldehydes and alcohols formed are predominantly the "expected",
unrearranged products (Table 4).

The contrast between iodine and iodide is more apparent when a

primary alcohol, I-propanol, is homologated. Using iodine at IICo =
8.5, the rate of propanol homologation is slow and the ratio of
homologated product to olefin is only 0.12 (Table 5). The rate of
homologation is about 3.7 times faster when NaI is the activator and
the ratio of homologated product to olefin is 2.5 (Table 4).

Clearly a competition exists between an El mechanism favoring

tertiary alcohols and an SN2 process favoring primary alcohols
(Scheme 1). Both pathways are simultaneously operative. In the ab-
sence of an appreciable concentration of a strong nucleophile, pri-
Table 5. Carbonylation of C3-C4 Alcohols with Cobalt Carbonyl/Iodine Catalyst,a mmol of Products :t>
Carbonylation m
Alcohol + 5% Water Aldehydes Alcohols Hydrocarbons Ethers Others Hydrocarbons en
n-PrOH Butanal = 18 I-Butanol = 9 C3's 330 95 30 0.12 o
2-Methyl- 2-Methyl-l- J:
propanal = 9 propanol = 4
t-BuOH 2,2-Dimethy1- 2,2-Dimethy1- C4's 764 36 9 0.05 o
propanal = 3 propanol = 0 s:
3-Methyl- 3"';'Methyl-l- oG)
but anal = 14 butanol = 19
2.26 -t
2,2-Dimethy1- 2,2-Dimethyl- C4'S 236 17 15 5
propanal = 45 propanol = 7 z
t-BuOHb 3-Methyl- 3-Methyl-l-
butanal = 458 butanol = 23

aCatalyst = Cobalt Carbonyl = 0.5 mmol, 12 4.25; Alcohols 95 cc

Pressure = 4000 psig, H2/CO = 1
Temperature = 200°C
Time = 3 hr
bI2 = 0.25 mmol

mary alcohols will not react since the SNZ path is minimized and
elimination is not favorable for primary alcohols.

-HZO -w
R+ - - - -__~ olefin co + HZ _ RCHZOH \

El pathway HCo(CO)4
ROH + ~ ~ [ROHZ]+

----- Z
-HZO [Co]

Scheme 1. Carbonylation Pathways

When the concentration of nucleophile is high, the SNZ mechanism

can proceed at an appreciable rate for primary alcohols. In this
case, the El reaction can still occur for tertiary alcohols. How-
ever, the intermediate olefins now accumulate since the concentration
of effective hydroformylation catalyst has been significantly reduced
by the large amount of iodide and phosphine present. The ratio of
homologated product to olefin is shown in Table 4. Considering the
above arguments, this ratio may also be considered to reflect the
ratio of SNZ to El reactions. Accordingly, this ratio is expected
to be large for primary alcohols, and become progressively smaller
for secondary and tertiary alcohols. This is, in fact, observed.
Primary alcohols show ratios of 1.Z to 3.9, while for secondary al-
cohols the ratios range 0.4 to 1.8. The ratio of homologated product
to olefin is only 0.03 for t-butanol.


Unpromoted Cobalt Systems

Higher alcohol homologation undoubtedly involves alcohol dehy-

dration to olefin catalyzed by the strong acid, HCo(CO)4. Olefin
hydroformylation catalyzed by HCo(CO)4 then leads to the observed
products. The special case of methanol, which cannot be dehydrated
to an olefin, probably involves initial formation of a protonated
methanol and subsequent nucleophilic attack by Co(CO)4 to give
CH3Co(CO)4. Methyl migration gives an acetyl intermediate, which is
reduced by either HZ (intramolecular) or HCo(CO)4 (intermolecular)
to acetaldehyde.

Promoted Cobalt Systems


A strong, covalent activator like iodine, promotes higher alco-

hol dehydration to olefin' (most likely via HI). When I/Co < 2, this
olefin is hydroformylated by HCo(CO)4. When I/Co > 2, olefin is
formed at a rate greatly exceeding the rate of hydroformylation.
Little HCo(CO)4 is present under these conditions and hydroformyla-
tion is suppressed. We believe that HCo(CO)4 is effectively removed
by formation, under reaction conditions (125 atm CO, 125 atm H2), of
(Co(I)n+l(CO)x)-n species. Hydridocobalt tetracarbonyl is in equi-
librium with H2 and dicobalt octacarbonyl, which disproportionates in
alcohols to CoT2 and Co(CO)4. This disproportionation is catalyzed
by iodide. The logical anion of Co+2 is I-. An unstable Co(I)2(CO)
species has been reported to form when CoI2 is subjected to high CO
pressure [23]. Further, (CoI2(CO)(PR3)2) complexes have been shown
to be reduced to (CoI(CO)2(PR3)2) species in the presence of CO and
excess phosphine [24]. It is not unreasonable to propose that at
elevated temperature and pressure, hydrogen can replace phosphine as
a reducing agent, leading to Co(I) carbonyliodide complexes. We
postulate that such complexes cannot effectively activate olefins for
the hydroformylation reaction.

With respect to methanol homologation using a strong activator

with I/cb < 2, methyliodide is formed, which is subject to attack by
Co(CO)4 to give CH3CO(CO)4. As the I/Co ratio increases, it is pos-
sible that methyl iodide oxidatively adds to a (Co(I)n+l(CO)x)-n spe-
cies. Recall that when I/Co > 2, ethanol production is suddenly
suppressed. We believe that aldehydes are reduced to alcohols by
HCo(CO)4 and when I/Co > 2, little of this reducing catalyst is pres-
ent (vide supra).

Phosphine analogs of such cobalt intermediates can be readily

envisioned for those systems containing such modifiers. When iodides
are used as promoters, higher alcohol homologation proceeds through
alkyliodide intermediates. We propose that such intermediates oxi-
datively add to cobalt species described above.

It is interesting to note that both strong and mild activators

may be employed together with interesting results. Recently, it was
disclosed that use of covalent and ionic iodide promoters in combina-
tion is particularly effective [25,26]. Generally, the ionic form
of iodide is employed in excess, relative to the covalent form, and
both forms are in excess relative to cobalt. A 2- to 4-fold rate
enhancement was observed relative to the ionic and covalent iodides
acting independently, suggesting a synergistic effect.


We thank Drs. T. V. Harris and D. C. Hrncir; and Mr. R. Bartek

for their inspiring discussions, and Mr. S. D. Gourley for his tech-
nical assistance. We also thank Gulf Research & Development Company

for support and permission to publish this work.


l. G. Wietzel, et al., German Patent 867,849 (1941).

2. G. Wietzel, o. Vorbach, and A. Scheuermann, German Patent
875,346 (1953).
3. 1. Wender, R. Levine, and M. Orchin, J. Am. Chern. Soc., 71:4160
4. D. W. Slocum, in "Catalysis in Organic Chemistry," W. H. Jones,
ed., Academic Press, Inc., New York, pp 245-276 (1980).
5. F. Piacenti and M. Bianchi, in "Organic Synthesis via Metal Car-
bonyls," 1. Wender and P. Pino, eds., Vol. 2, John Wiley &
Sons, New York, pp 13-18 (1977).
6. H. Bahrmann and B. Corni1s, in "New Syntheses with Carbon Mon-
oxide," J. Falbe, ed., Springer-Verlag, New York, pp 226-242
7. 1. Wender, R. A. Friedel, and M. Orchin, Science, 113:206 (1951).
8. K. H. Ziesecke, Brennstoff-Chem., 33:385 (1952).
9. W. F. Gresham, U.S. Patent 2,623,906 (1952); assigned to E. 1.
du Pont de Nemours & Co.
10. J. Berty, L. Marko, and D. Ka110, Chern. Techn., 8:260 (1956).
ll. A. D. Riley and W. O. Bell, U.S. Patent 3,248,432 (1966);
assigned to Commercial Solvents Corp.
12. G. N. Butter, U.S. Patent 3,285,948 (1966); assigned to Commer-
cial Solvents Corp.
13. M. Kuraishi, S. Asano, and A. Takahashi, U.S. Patent 3,356,734
(1967); assigned to Commercial Solvents Corp.
14. M. Kuraishi, S. Asano, and Y. Shinozaki, U.s. Patent 3,387,043
(1968); assigned to Commercial Solvents Corp.
15. K. H. Keirn, J. Korff, W. Keirn, and M. Roper, Erdoe1 Koh1e,
Erdgas, Petrochem. Brennst-Chem., 35:297 (1982).
16. W. W. Paudler and T. E. Walton, J. Org. Chern., 46:4306 (1981).
17. I. Abis, A. Sen, and J. Halpern, J. Am. Chern. Soc., 100:2915
18. D. J. Drury, M. J. Green, D. J. M. Ray, and A. J. Stevenson,
J. Organometal. Chern., 236:C23 (1982).
19. W. R. Pretzer and T. P. Kobylinski, Ann. NY Acad. Sci., 333:58
(1980) •
20. a. W. R. Pretzer, T. P. Kobylinski, and J. E. Bozik, U.S. Patent
4,133,966 (1979); assigned to Gulf Research & Development Co.
b. W. R. Pretzer, T. P. Kobylinski, and J. E. Bozik, U.S. Patent
4,239.924 (1980); assigned to Gulf Research & Development Co.
c. R. A. Fiato, U.S. Patent 4,233,466 (1980); assigned to Union
Carbide Corp.
d. B. R. Gane and D. G. Stewart, U.S. Patent 4,262,154 (1981);
assigned to The British Petroleum Co.
e. W. R. Pretzer, T. P. Kobylinski, and J. E. Bozik, U.S. Patent
4,346,020 (1982); assigned to Gulf Research & Development Co.

f. M. M. Habib and W. R. Pretzer, U.S. Patent 4,352,947 (1982);

assigned to Gulf Research & Development Co.
21. G. R. Burns, J. Am. Chern. Soc., 77:6615 (1955).
22. w. Corne1y and B. Fell, J. Mol. Cata1., 16:89 (1982).
23. w. Hieber and H. Schulten, z. anorg. u. a11gern. Chern., 243:145
(1939) •
24. M. Basato, M. Bressan, and P. Rigo, J. Organometa1. Chern., 232:
81 (1982).
25. J. Gauthier-Lafaye and R. Perron, U.S. Patent 4,306,091 (1981);
assigned to Rhone-Pou1enc Industries.
26. J. Gauthier-Lafaye and R. Perron, U.S. Patent 4,324,927 (1982);
assigned to Rhone-Pou1enc Industries.


M. L. Deem

Department of Chemistry
Lehigh University
Bethlehem, Pennsylvania 18015



Carbon monoxide and hydrogen (1:1) in the presence of various

metal carbony1s (of Co, Rh, Ru, Mn, Re, V, Cr, and Os) give ethanol,
predominantly, at >200°C and >1000 Torr. Inclusion of po1yethers in
these homogeneous catalytic systems allows meaningful conversions to
products at <200°C and <200 Torr, and the highest rates of reaction
are observed with the cobalt and ruthenium catalysts [1]. Product
analyses and chemical analogies have been used to decipher the spe-
cial function of the po1yethers. These additives solvate isolates
({(dig1yme)nCo[Co(CO)4]3}, {[(dig1yme)2Co]2+[C06(CO)15]2-}) from the
reactors. Breakdown of the po1yethers with incorporation of them into
products (e.g. CH3COCH3 and CH30COH) also is found [1]. Direct at-
tachment of po1yether moieties to the cobalt, e.g. in 1 [1], is indi-
cated both by infrared (IR) spectroscopy and by analogy with known

oII 0
[.0\ ~ \ }; /0:\
\.0 'oj

organometallic chemistry, e.g. in LnCo-0(CH2CH20)Me and H-CoLn,CH2-

(OCH2CH2)20Me [1]. This latter structure is formed upon insertion
of cobalt into a carbon-to-hydrogen bond.

Carbon monoxide fixation occurs in other homogeneous-phas~

catalytic systems, and hydroformy1ation and acetic acid-from-methano1
288 M. L. DEEM

are industrially significant examples, which are catalyzed by rhodi-

um(I). An oxidative addition to Rh(I) is a chemical step in these
processes. Oxygen-containing compounds (i.e. ethers and alcohols)
can influence this step, a representative example of which can be
directly followed by nuclear magnetic resonance (NMR) spectroscopy
as the reaction proceeds in the laboratory. Some of these molecular
interactions at a rhodium site are considered below.



Binding of organorhodium compounds to solvents is firmly estab-

lished by elemental analysis (cf. Table 1) and X-ray crystallography.
Solvation of the Rh(I) compound RhX(CO) [PhZPCH2CH2(OCH2CH2)nPPhZ]
(where n = 1 - 3 and X = Cl, PF6) with oxygen atoms of EtOH [6], HZO
[7,8], and the f-CH2CH20~ group [7,8] of the ligand has been studied
by X-ray crystallography. Bound solvents show infrared and nuclear
magnetic resonance bands, which are shifted from bands of the free
solvents. Although not isolable, unstable, loosely associated,
and dynamically exchanging solvates all have been observed spectro-
scopically [9].


Solvents effect the rate of oxidative addition to Rh(I). Two

concepts for this are the "outer sphere mechanism" [10] and a mech-
anism in which solvent participates as a displaceable ligand. This
latter mechanism (equation 1) is kinetically supported [ll,lZ] for
the hydrogenation of alkenes. Strongly coordinated solvents block

ll~s alkene,

+ alkene11 ~e 11+ alkene, -S

RhIX(S) (alkene)LZ +~ HZRhIIIX(alkene)LZ

incorporation of either the alkene or hydrogen into a reaction inter-

mediate (equation 1), thereby inhibiting hydrogenation. In the outer-
sphere mechanism, solvent in the outer coordination sphere of the
metallic ion holds a coreactant in close association with the metal-
lic ion until electron transfer finalizes the reaction. In this
mechanism, coulombic and entropic terms become especially important.

Table 1. Elementally Analyzed Solvates of Some

Organorhodium Compounds

Solvate Reference

RhC12(COC2HS) (PPh3)2·C2HSOH 2
RhCl(PPh3)2(dioxane) 3
RhC12(COC2HS) (PPh3)2·0.SCH2C12 2
[H2RhCl(PPh3)2]2·CH2C12 4
[H2RhBr(PPh3)2]2·CH2C12 4
[H2RhCl(PPh3) (CSHSN)]2 4
[H2Rh(PPh2cy) 2 (Me2CO)2]PF6a S
[H2Rh(PPh2cy) 2 (MeCN)2]PF6 S

a cy = cyclohexyl.

A composite [12-16] order of activity for solvent effects on

alkene hydrogenation catalyzed by RhCl(PPh3)3 is PhN02 > MeCOEt,
Me2CO > other PhZ > all haloorganics > MeCN. Last-ranked acetoni-
trile is inhibitory. Varied, but decided, influences on the rate of
this alkene hydrogenation have been reported for oxygenated solvents
[for alcohols [IS], (MeOH [4,14], EtOH [4,12,14,16,17], tert-BuOH
[4], and iso-PrOH [4]) and for ethers [IS], (PhOMe [16], tetrahydro-
furan [14,16], and dioxane [4])]. Utilization of one of these alco-
hols as a cosolvent with benzene accelerates the hydrogenation re-
Added alcohols and ethers also influence the optical yields from
asymmetric hydrogenation. In one system [18], change of solvent from
MeOH to iso-PrOH enhanced (by a small amount) the optical yields from
a representative ketone, imine, and alkene.

Based on mechanistic considerations, a requirement for only

stoichiometric amounts of an oxygen-containing promoter is antici-
pated. Incorporation of an o-anisyl group (2-MeOC6H4-' which can be
considered to be an internal solvent that has its oxygen atom bound
to rhodium) onto a phosphine-terminated ligand chain provides an ef-
fective chiral ligand [19,20]. An o-anisylphosphine ligand imparts
a positive effect on a different oxidative addition. The rate for
oxidative addition of alkyl halides with RhCl(CO) (PMe2Ar ) 2 is 100-2S0
fold faster in its presence (as compared to runs with its para
analog) [21].
290 M. L. DEEM


Surfactants facilitate some rhodium-catalyzed reactions. His-

torically, surfactants have been added to achieve reactions with
rhodium in partially aqueous media.

The ligand Ph2PC6H4-m-S03Na is used in RhC1(PAr3)3 [22,23] and

in HRh(CO) (PAr3) 3 [23,24] to permit two-phase (aqueous-organic)
hydrogenation and hydroformy1ation of a1kenes. This ligand provides
solubilization and, hence, reaction of the rhodium(I) in water. With
tetra-n-hexy1ammonium bisulfate as the phase-transfer agent, transfer
hydrogeno1ysis of aryl bromides is accomplished [25] in the organic,
rhodium-containing phase with formate ion as the hydrogen donor.

Amides of bis(2-dipheny1phosphinoethy1)amine, which bear water

solubilizing groups (-C02-, -S03-, f-CH2CH20-1), coordinate through
the phosphorous atoms with Rh(t). Use of these catalysts permits
[26,27] hydrogenation of a test alkene, 2-acetamidopropenoic acid,
in partially aqueous (dioxane-water), miscible solutions. Generally,
catalytic activity directly correlates with the water-solubilizing
ability of the ligands. An exception is the (impure) ligand CH3(OCH2-
CH2)nCO[N(CH2CH2PPh2>21, with n = 12, 16, 110, which is stated [27]
to be the best solubi1izer but a poor activator for Rh(I) catalysis.
Another ligand {H02C(CH2)2CO[N(CH2CH2PPh2)21} provides useful cata-
lytic activity in aqueous solutions only if a low level of the
micelle-former sodium dodecy1 sulfate [27] also is present. Other
ligands, e.g. Ph2P(CH2)nC02- with n = 2,4,6,8,10, function with Rh(I)
in micellar aqueous systems to effect [28] hydrogenation of a1kenes.

In most of the reports, which are summarized above, the surfac-

tants serve to solubilize reactants in partially aqueous solvent
systems. The groups -S03- and -C02- have been incorporated (gener-
ally in stoichiometric amount) into phosphine ligands on Rh(I) to
carry the rhodium into water in phase-transfer reactions, to solubil-
ize the rhodium in mixed water-dioxane, and to form reactive micellar

Quaternary ammonium compounds have served a) to carry a coreact-

ant (HC02-) from water into a rhodium-containing phase (to hydrogen-
olyze ArBr) and b) to solubilize the iodide ion in an anhydrous system
(of RhC1(CO)(PR3)2 and Mel). Tetra-n-propy1ammonium iodide functions
[29] as a solubilized source of iodide ion in the neat system of
RhC1(CO) (PR3) 2 and Mel (Mel also was the solvent). An increasing
order of reactivity in the oxidative-addition reaction of RhC1(CO)-
(PR3)2 and Mel is observed [29] for the ligands PPh3 < P(C6H4C6H13-
P)3 < P(C6H4C2H5-p)3, P(C6H4C4H9-p)3 < P(n-C18H37)3 < P(n-C4H9)3 <
P(n-C8H17)3. These kinetic differences are attributed [29] to a com-
bination of electronic and steric factors for these stoichiometric
levels of alkyl-substituted ligands. Solvent characteristics of the
ligands are of interest, but this influence on the RhC1(CO) (PR3)2/MeI

reaction remains obscure.

Addition of a neutral polyether surfactant to a representative

oxidative addition of Rh(I) accelerates [30] the addition. The mag-
nitude of enhancement is comparable to that achieved upon change of
solvent, although small (less-than-stoichiometric) amounts of a sur-
factant are added. The study system consists of either methoxy(eth-
yleneoxY)10hexadecane [31] or diphenylphosphino(ethyleneoxyhohexa-
decane [32] with propanoyl chloride and chlorotris(triphenylphos-
phine)rhodium(I) in CDC13 (equation 2).
ClCOC2H5 + RhICl(PPh3)3 -----. cis- and trans- (2)
- PPh3 RhC12(COC2H5) (PPh3) 2

The cis and trans products are differentiable [30] in lH and

3lp{lH} NMR spectra (Table 2). In the standard (surfactant-free) re-
action system, a variable temperature NMR study (cf. the experimental
section) has revealed that the reaction of ClCOC2H5 is not diffusion-
controlled. Unreacted ClCOC2H5 is observed [30] in the lH NMR spec-
tra after one to two min of reaction at room temperature. Stepwise
temperature increases (cf. Table 3) for this reaction mixture, which
had been quenched in liquid nitrogen after 1-2 min, in an NMR probe
show that the cis isomer is the kinetically-favored and the trans
isomer is the thermodynamically-favored product. Reaction kinetics
have been measured [30] for a related system. The efficacy of D20
as a quenching agent was verified using lH NMR spectroscopy (no back-
reaction or hydrolysis of RhC12(COC2H5)(PPh3)2 was observed), and
D02CC2H5 (the product from hydrolysis of ClCOC2H5) was measured by
gas chromatography (cf. Table 4). An induction period exists, in
which second-order reaction kinetics are not rigorously followed.
With the assumption of second-order reaction kinetics after the in-
duction period, kinetic constants were estimated to be increased
five- to eight-fold in systems to which the polyethers have been

Surfactants, in general, assume somewhat-ordered configurations

and aggregate to modify molecular organizations at reactive sites.
In this way, electron density at Rh(I) is enhanced in tandem with the
tendency [33] for oxidative addition. Dipolar interactions between
both the rhodium center and the carbonyl group of ClCOC2H5 with the
ethereal oxygen atoms hold these reactants in close contact until the
propitious moment for reaction exists (compare this to the "outer
sphere mechanism" for solvent activation of organometallic reactions
in the previous section). Higher local concentration gradients of the
reactants exist in the millieu of surfactants as compared to the bulk
phase, and this has a reaction-promoting effect. To maximize binding
of reactants to polyethers, water should be rigorously excluded from
these reactions. This is in decided contrast to micellar and phase-
transfer situations, which are aqueous or partially aqueous systems.
292 M. L. DEEM

Table 2. NMR Signals for cis- and trans-RhC12(COC2HS)(PPh3)2

1H NMRa 31p{lH} NMR

o (ppm) 0 (ppm) 0 J (Hz)

cis-RhC12 (COC2HS) (PPh3) 2 4.04, 1.19-1.17 30.8 lS0

trans-RhC12 (COC2HS) (PPh3) 2 2.96, 0.78 23.8 109

aFrom Reference 2.
bDownfie1d to 8S% H3P04 is positive.

Any water in the system will preferentially bind to the deliquescent

po1yethers and, thereby, destroy alternative binding opportunities.



Solvents and surfactants bind to organorhodium compounds, and

in the absence of more competitive reaction partners, the solvents
and surfactants themselves can react with the rhodium ion. An exam-
ple of this is rhodium-catalyzed transfer hydrogenation. Transfer
hydrogenation is a redox reaction in which a donor molecule gives up
two atoms of hydrogen to an acceptor molecule. Both the donor and
the acceptor complex with Rh(I). This couple is shown in equation 3
for an alcohol (the donor) and an alkene (the acceptor).
R,R'CHOH + ~C=C~ ~ R,R'C=O + HC-CH (3)
Itdonor" "acceptor"

Dehydrogenation (equation 3) does not proceed, in some instances,

without a particular hydrogen acceptor. Other dehydrogenations are
more facile (e.g. of 1,2-dipheny1ethane [34]), and the fate of the
lost hydrogen remains unreported. The earliest evidence for dehydro-
genation of alcohols by Rh(I) is in reports [4,3S-37] of incorpora-
tion of carbon monoxide, which is generated from alcohols (equation
4), into the rhodium(I) complex.
RCH20H + RhC1(PR'3)3 - HRhC1(PR'3)2(COR)
-PR' 3 +
RhC1(CO) (PR'3)2 + R-H (4)

These sequential reactions involve binding of a primary alcohol to

Rh(I), transfer of two atoms of hydrogen from the alcohol to rhodium

Table 3. Isomeric Composition of RhC12(COC2H5)(PPh3)2

upon Temperature Increase

Temperature (DC) cis/trans

-55 100/0
-30 75/25
-15 66/33
o 45/55
+15 40/60
+30 20/80

with formation of an aldehyde, and decarbonylation of the aldehyde

with formation of RhCl(CO)(PR'3)2 (equation 4). This proceeds most
readily for allylic alcohols [35-37] or with other unsaturated pri-
mary alcohols [38], which advantageously complex to Rh(I) via their
olefinic IT-bond.

Hydrogen atoms, which are given up by primary or (somewhat less

rapidly) secondary alcohols, are captured by appropriate acceptors
to transfer hydrogenate (See Table 5). The alcohols serve as hydro-
gen-donors (and sometimes as the reaction solvent) with a1kenes,
alkynes, 1,3-dienes, other dienes, a1dimines, ketones, and a,a-unsat-
urated ketones. Dienes react more facilely than do monoenes, and
the olefinic bond of a,a-unsaturated ketones can be [43] preferen-
tially reduced. The transfer of hydrogen from an alcohol to the ole-
finic link is believed to be concerted, based on the regiospecific
incorporation of deuterium and hydrogen into an unsaturated ketone
[47]. Cleavage of the C-H bond is [39,47] the rate-determining step
in these transfer hydrogenations at Rh(I).

Although different mechanistically, transfer hydrogenation

occurs with ethers as the hydrogen donors. High temperatures (>100DC
[41]) have been required. From among the ethers, dioxane exhibits
[49] a decidedly superior activity with olefins (as compared to 1,2-
dimethoxyethane, l,2-diethoxyethane, l,2-diacetatoethane, 2,2-diac-
etatoethane, l-methoxy-2(2-methoxyethoxy)ethane, 1,2-dimethoxycyclo-
hexane, and oxane). Displacement by dioxane of a phosphine ligand
in chlorotris(triphenylphosphine)rhodium(I) occurs [3] at ambient
temperatures (cf. the elemental analysis in Table 1) to exhibit [3]
two infrared bands, which are appropriate for the coordinated·diox-
ane. Kinetic isotope studies with hydrogen- and deuterium-labelled
dioxanes have established that the rate-determining step in coupled
transfer hydrogenation is loss of two atoms of hydrogen from the
dioxane [3J. Most a1kenes take up the hydrogen at comparable rates;
294 M. L. DEEM

Table 4. Reaction of C1COC2H5 with RhCl(PPh3)3a

System Time (min) Consumption of C1COC2H5 (%)

Additive-free, 1.20 8.9

33.7 ± 106°C 3.125 47.7
MeO(CH2CH20)10C16H33 0.67 72.7
present [31],
29.8 ± 0.4°C 1.52 86.8
Ph2P(CH2CH20)lOC16H33 0.62 85.3
present [32],
29.20 ± 0.02°C 2.13 87.5

aReactant ratios are [RhCl(PPh3)3]:[surfactant]:[C1COC2H5]:[D20]


complexation between the rhodium and alkene is indicated [3] as oc-

curring after the rate-determining reaction step of hydrogen transfer
from dioxane to Rh(I).


Alcohols and ethers bind to organorhodium compounds, promote

oxidative additions to Rh(I), and participate as a coreactant in a
special class of oxidative additions (transfer hydrogenations at
Rh(I» as described in the preceding sections. The ability of ethers
to bind to Rh(I), to serve as carriers of associated alkyl chains,
and to hold alkyl carbon-to-hydrogen bonds in close proximity to
Rh(I) for their subsequent insertion by rhodium has been tested. A
chronology of reaction events will be described.

Dynamic events in the system Ph2P(CH2CH20)10(CH2)15CH3/RhCl-

(PPh3)3/RhC12(COCH2CH3)(PPh3)2 (1.0:0.2:1.0, based on moles in CDC13
in the dark at room temperature and atmospheric pressure) become vis-
ible in NMR spectra. Ongoing exchange of alkylphosphine 6 for a
triphenylphosphine ligand (~) of RhCl(PPh3)3 (Z) is shown-[30] in
Ph2P (CH2CH20) 10C16H33 + Rh I C1( PPh 3)3
6 7
l (5)
+ RhICl (PPh3) 2 [Ph2P(CH2CH20)10(CH2)15CH3]
~ 9

Table 5. Transfer Hydrogenation with Alcohols;

R,R'CHOH + X=Z ~ R,R'C=O + HX-YH
"donor" "acceptor"
2 3 4 5

Alcohol/Donor, .k Acceptor, ,;t Temper- Yield (i.) Refer-

R R' x Y ature 2 ence

H H 80 39

H H 80 39

H H CHCH:CH 2 80 39

H H CHCCH 3 :CH 2 60, 80 (mixed 39


H H CHCH:CHCH 3 80 39

H CH 3 CHCCH 3 :CH 2 80 39

H CH 2CH 3 CHCCH 3 :CH 2 80 39

CH 3 CH(CH 2 ) 4CH 3 83 100 40

CH 3 CH(CH 2 ) 5CH 3 20, 100 5-6 41

CH 3 CHPh 83 29 40

CH 3 80, 83 39, 40

CH 3 norbornene 42

CH 3 norbornadiene 83 40

CH 3 1,5-cyclooctadiene 83 40, 42

CHiH3 ~H(CH2)5~H 80 42

CH 3 ~H(CH2)iH 180 (some) 41

II 1\
CH 3 CH(CH 2 ) 4CH 83 83 40-42

CH 3 "
2 \CH 100 38 42, 43

CH 3 ~H(CH2\~H 42

CH 2CH 3 3H(CH2)5~H 80 42

(CH 2 ) i H 3 3H(CH 2 ) 5~H 80 42

3-cyclo- CH(CH 2 ) 4CH 145-155 38
H Ph CH(CH 2 \CH 40 43
H Ph CH(CH 2 )5CH 80 42
Ph CH(CH 2 )4 CH 43

Ph ~H(CH2) 4~CH3 43
296 M. L. DEEM

Table 50 (continued)
Alcoho17Donor! ~ Acceptor! ..:L Temper- Yield (%) Refer-
R R' X Y ature 1- ence

H PhCH:CH HZC CHPh 145-155 76 38

H PhCH:CCH 3 H3CCH CHPh 145-155 67 38

H PhC!C HZC CHPh 145-155 "loW" 38

H PhC!C HC CPh 145-155 "loW" 38

CH 3 CH 3 0 CCH 3,(CH Z)3CH3 83 97 44

CH 3 CH 3 0 C(CH Z)4 45

CH 3 CH 3 0 C(CH Z\ 80, 83 90, 9 44, 46

CH 3 CH ZCH 3 0 C(CH Z)5 45

CH 3 CH 3 0 C(CH Z)4(CH,CH 3 ) 83 50S 44

CH 3 CH 3 0 C(CH Z)(CH,CH 3 )(CH Z)3 83 30 44

t 83 98, 99 44, 46
CH 3 CH 3 0 C(CHZ)Z(CH, Bu)(CHZ)Z

CH 3 CH 3 0 CCH 3,(CH 2)ZCH:CH Z 83 8a , 305 b 44

CH 3 CH 3 0 CCH3,(CH2)ZCCH3:CHZ 83 Z005 c , Z8 b 44

CH 3 CH 3 o. CCH 3,(CH Z)ZCH:C(CH 3)Z 83 Sa, 15 b 44

CH 3 CH 3 gH(CO)(CH2)3~H 40 15 43

H Ph ~H(CO)(CHZ)3~H 40 90 43

CH 3 Ph ~H(CO)(CH2)3~CH3 70 43

CH 3 Ph ~H( CO)( CH Z) 3~H 40 47

CH 3 Ph HZC CHCOCH 3 40 43

CH 3 Ph CH 3CH CHCOCH 3 50 43
CH 3 Ph CH(CO)(CH 2)3CH 40, 50 46, -- 43

H CH 3 PhCH CHCOCH 3 50 Z7 43

CH 3 CH 3 PhCH CHCOCH 3 50 lZ 43

H 4-MeOPh CH(CO)(CH 2)3~ H

40 96 43

CH 3CH 2 Ph ~H(CO)(CHZ)3~H 40 Z9 43

CH(CH 3 )2 Ph ~H(CO)(CH2)iH 40 ZOoS 43

C(CH 3)3 Ph ~H(CO)(CHZ)3~H 40 4.5 43

n I
CH 3 CHlh CH(CO)(CH Z)3CH 40 18 43
, q
Ph Ph CH(CO)(CH Z)3CH 40 49 43

Ph CHlh lH( CO) (CH Z) 3~H 40 35 43

• n
Ph (CHZ)ZPh CH(CO)(CH 2)3CH 40 18 43

Table 5. (continued)

Alcohol/Donor! ~ Acceptor! .:L Temper- Yield (%) Refer-

R R' X Y ature 5 ence

H Ph PhCH CHCOCH 3 50 79 43

H 4-MeOPh PhCH CHCOCH 3 50 67 43

H 4-0 2NPh PhCH CHCOCH 3 50 none 43

CH 3 Ph PhCH CHCOCH 3 19, 50, 92, 23, 43

170 30

CH(CH 3)2 Ph PhCH CHCOCH 3 50 27 43

C(CH 3 )3 Ph PhCH CHCOCH 3 50 8 43

CH 3 Ph 4-0 2NPhCH CHCOCH 3 50 43, 47

CH 3 Ph 4-Me 2NPhCH CHCOCH 3 50 43, 47

CH 3 Ph PhCH CHCOC(CH 3 )3 50 43

CH 3 Ph ArCH CHCOPh 50 43

Ph Ph PhCH CHCOPh 50 39 43

Ph CH 2Ph PhCH CHCOCH 3 50 28 43

4-MeOPh CH 2Ph PhCH CHCOCH 3 50 25 43

CH 3 CH 3 PhCHl CHPh 83 91 48

CH 3 CH 3 PhCH 2N CHAr 83 81-93 48

CH 3 CH 3 PhN CHPh 83 85 48

CH 3 CH 3 ArN CHPh 83 78-96 48

CH 3 CH 3 PhN CHAr 83 78-96 48

aFor ketone.
bFor alcohol(s).
cFor alkane.

3l p {lH}FT NMR spectra as broadened signals at 0 43.24, -36 - 31, and

-3.18 to -3.96. These are characteristic of a fluxional trans Cl-Rh-
P array in a planar four-coordinate Rh(I) molecule [50-52], rhodium-
coordinated phosphine, and PPh3 [50]. A downfield signal simultane-
ously appears at 0 -88. After ten days, these broad bands disappear
(0 43.24 and -36 - 31) or stabilize [0 -3 to -4 at 0 -5.57 (sharp)
and 0 -88 at 88.00-87.95 (d-d, JRhP = 112 - 114 Hz, Jpp = 17 - 19
Hz)]. This doublet-of-doublets shows a 3lp chemical shift value com-
patible [53-55] with literature reports for constrained rhodium-com-
plexed phosphines. The magnitudes of the coupling constants associ-
ated with 0 88.00 - 87.95 are best fit to known 3lp NMR numbers for
octahedral Rh(III) complexes (cf. Table 6). Characterization of this

Table 6. J RhP and Jpp for Complexes of Rh(I) and Rh(III)

Coordination JRhP (Hz)

Number of Rh J pp (Hz)
Equatorial Axial

4 189-192 for 37-38 [51]

trans C1-Rh-P [5~]
124-146 for
trans P-Rh-P [51,52]
5 145-146 [54] 111-116 [54] 46-70 [50,54,56]
6 92-114.for 112-116 for 17-33 [51,52,58,59]
trans P-Rh-P [51,57] trans C1-Rh-P [57]
90-94 for
trans H-Rh-P [51,58]


constrained, octahedral Rh(III) complex as H-Rh-CH(CH2-)2 (11, equa-

tion 6) is supported by observed l3C{lH} NMR signals [30] and by the

: CH 2- -----7'"-=------.... -CH2CHCH2CH2CH2-
RhlLu ~h Ln
10 11

.. -CH 2CH=CHCH 2 - (6)
RhIL n
+ RhIL n 12
Z oxygen of an ether

presence of the transfer hydrogenation products (both 12 and 13) of

11. A signal for H-Rh is not observed [30,55] at room-temper;ture.

The coordinated alkene 13 is observed in the above reaction mix-

ture within 24 hr. A lH NMR ~ignal 10
4.50 (br, coalesced free and
bound olefinic CH [55,60,61])] and 3C{lH}FT NMR signals {o 61.67 -
61.65 (d, J = 11 Hz; (-CH2CB:-)2RhILn [55,60,62]) and 17.82 (t, J =
12 Hz; (-CB2CH:-)2RhILn [55,60,62])} corroborate the formation of 13.
These results further establish that rhodium has been inserted into-
a C-H bond on an alkyl chain and at a site remote from a rhodium-
coordinated phosphorous atom of a monodentate phosphine. Oxidation
of !! to 12 is coupled with reduction of RhC12(COC2H5)(PPh3)2, which
is formed from ClCOC2H5 and RhCl(PPh3)3 (equation 2), to n-PrOH.
Confirmatory NMR bands (for n-PrOH) in the reaction mixture are com-
pared to bands of known n-PrOH in Tables 7 and 8.

Hydrogen transfer to cyclohexanone, rather than to propanoyl

chloride, failed in the system of Ph2P(CH2CH20)10(CH2)15CH3 (1.0
mol):RhCl(PPh3)3 (0.1 mol):cyclohexanone (1.1 mol) under room condi-
tions [30].

For the sequence lQ + 1~ (equation 6), the alkyl chain can

merely be complexed by an ether group to rhodium(I). A phosphine
ligand is not required. A mixture of MeO(CH2CH20)10(CH2)15CH3:RhCl-
(PPh3)3 (= 0.5 mol:l.l mol) readily yields a rhodium-coordinated
alkene [IH NMR (CDC13, 60 MHz) 0 4.26 (br)] at room temperature and
pressure. Subsequent addition of ClCOC2H5 (1.0 mol) leads to RhC12-
(COC2H5) (PPh3}2· Upon completion of this acylation, injection of D20
causes a progressive shift of the lH NMR coalescence band from 0 4.26
(broad) to 0 4.67 (broad) (Table 9). The D20 associates with the
polyether and removes it from the locus of the rhodium. Some Rh(I)-
olefinic complexation is broken, with a resultant downfield shift of
the coalesced lH NMR band (for the combined free and rhodium-bound
300 M.L.DEEM
Table 7. Characteristic Bands of n-PrOH
in l3C{lH}FT NMR Spectra

o (ppm) Found
Reaction Mixture Known n-PrOH

64.413 64.522
10.266 10.185

alkenes) toward the value of the completely free alkene. The addi-
tion of D20 breaks polyether-rhodium(III) interactions and causes
two other simultaneous changes in the lH NMR spectrum. All residual
cis-RhC12(COC2H5)(PPh3)2 is converted to its trans isomer following
introduction of D20, where this transformation is monitored by the
highly characteristic lH NMR signals for the COCH2CH3 group of the
cis and trans isomers (cf. Table 2). This may imply that association
of MeO(CH2CH20)10(CH2)15CH3 with some cis-RhC12(COC2H5) (PPh3) 2 inhib-
ited the cis to trans conversion, possibly by impeding a Berry pseu-
dorotation process. The -CH20- units of MeO(CH2CH20)10(CH2)15CH3
show two characteristic signals. The dominant one is at 3.64-3.61
ppm, and an environmentally sensitive signal is present in the region
of 3.72-3.66 ppm. This latter signal shows variations in its rela-
tive intensity following successive addition to MeO(CH2CH20)10(CH2)15-
CH3 of RhCI(PPh3)3 plus CICOC2H5 and, then, D20. Finally, after the
mix is left with D20 for about one day, this minor band disappears.
More dramatic interactive effects are exhibited [30] in the IH NMR
spectra for a polyether following its admixture with R4~, which is
a less diffuse cation than are Rh(I) and Rh(III). A tight 1:1 ad-
duct, where the solid is stable after thermolysis at >65°C and 3-4
Torr for >3.75 hr, of l,2-dimethoxyethane and n-octyltriethylammonium
chloride shows lH NMR resonances of 3.23, 3.10, and 3.02 ppm (of rel-
ative intensities 4:2:3), whereas the separate components have res-
onances of 3.46, 3.53, and 3.36 ppm for (CH3CB2)4NCl, CH30C]2C]20CH3'
and C]30CH2CH20C]3, respectively.

Polyethers facilitate the insertion of Rh(I) into the alkyl

carbon-to-hydrogen bond at room temperature, atmospheric pressure,
and in the dark. Different approaches to the problem of insertion
by Rh(I) into the sp3 C-H bond have been pursued elsewhere. Photo-
activation of dihydrido(pentamethylcyclopentadienyl)trimethylphos-
phinerhodium(III) in propane at -55°C gives hydrido(pentamethylcyclo-
pentadienyl) propyl (trimethylphosphine) rhodium(III) [63]. Intramolec-
ular insertion of Rh(I) into nearby sp3 C-H bonds of a l,5-bisphos-

Table 8. Characteristic Bands of n-PrOH in lH NMR Spectra

Reaction Mixture Known n-PrOH

o (ppm) J (Hz) 0 (ppm) J (Hz)

~3.5 (interference) 3.58 (quartet)

2.38 (t)
--, --, 1.545, 1.468, 1.386, -- 1.58
0.97 7.33 0.93 (t) 7.32

i i i I
phinopentane complex has given [55,61] RhHC1(t-Bu2PCH2CH2CMeCH2CH2P-
Bu-t2) (in -30% yield) and RhHC1(t-Bu2PCH2CH2CHCH2CH2PBu-t2) (in 75%
yield). An X-ray crystallographic study [61] firmly establishes
these structures.

Greater reactivity toward metal insertion is shown by sp2 and

sp carbon-to-hydrogen bonds; the reactivity series for ease of metal
insertion is sp > sp2 > sp3 C-H bonds [64]. Stable acyl adducts have
been achieved upon oxidative addition of aldehydes to Rh(I), with
the result being equivalent to an addition across the aldehydic (sp2)
C-H link. Quinoline-8-carboxaldehyde readily reacts with chlorotris-
(triphenylphosphine)rhodium(I) to furnish an octahedral nitrogen-
complexed hydridoacylrhodium(III) compound (in 95% yield) [65].
Direct addition of acetaldehyde to chlorotris(trimethylphosphine)-
rhodium(I) furnishes [66] an octahedral hydridoacylrhodium(III) com-
pound with limited steric interactions and a resultant stability at
room temperature.

Selective functionalization of carbon-to-hydrogen bonds is the

goal of the carbon-to-hydrogen bond insertion work. A new realm of
synthetic organic methods can be envisioned for this chemistry. An
industrial process which might be based on an M + R1R2R3C-H + R1R2R3C-
M-H step is the conversion of methane to methanol, and this is of
industrial interest.


All NMR spectra were recorded with a JEOL FX90Q spectrometer,

which was equipped with a variable temperature probe. The exceptions
to this are the signals that are given in Table 9 and which were ob-
tained using a Hitachi Perkin-Elmer R-20A High Resolution NMR Spec-
trometer. The solvent was CDCI3' which contained Me4Si as an inter-
nal standard (0 = 0, with increasingly positive 13C and IH resonances
302 M. L. DEEM

Table 9. Position of the Coalesced lH NMR Bands

of !2 in the MeO(CH2CH20)10C16H33
Containing Reaction

Time, Following Addition

<5 (ppm)
of D20 (min)

4.26 (br) (zero)

4.31 (br) 8
4.41 (br) 15
4.52 (br) 30
4.67 (br) 50

to lower field). For 3lp NMR signals, 85% H3P04 was used as an ex-
ternal standard (0 = 0, with increasingly positive values to lower
field). For 3lp{lH}FT_, lH-, and l3C{lH}FT-NMR spectra, the probe
was set at 36.23, 89.55, and 22.49 MHz, respectively. Gas chromato-
graphy was done with a Hewlett-Packard 5730A chromatograph equipped
with a 5880A Series GC terminal. The single column measured 6' x 1/8"
(Ni) and was packed with 5% SE-30 on an 80/100 mesh support. Glass
injection port liners were loaded with silica gel (to prevent rhodium
from entering the column and injection port), and the injection port
and TC detector were maintained at 150 and 250°C, respectively.
Daily GC calibration curves were prepared using standard solutions
(equal to 13, 35, 55, and 78% reaction of ClCOC2H5)'


The conversion of synthesis gas to chemicals requires, in gen-

eral, reaction between molecular combinations of carbon, oxygen, and
hydrogen at a metallic center. Studies with rhodium compounds and
alcohols or ethers ("oxyhydrocarbons") are the specific focus of this

Existing literature shows that oxygen-containing organic com-

pounds interact beneficially with metallic rhodium-containing com-
pounds. These interactions lead to association complexes of varied
stabilities and with documented elemental analyses or spectral char-
acterization. Probably owing to this interactive tendency, oxyhydro-
carbons (which include alcohol and ether solvents and, also, some
surfactants) alter the course of chemical reactions between organic
and rhodium-containing compounds. An abundance of known homogeneous-
phase, rhodium-mediated transfer hydrogenations between alcohols or
ethers with hydrogen acceptors (such as alkenes) exemplifies the

synthetic potential of molecular interactions between oxygen and

rhodium centers.

Recent laboratory advances with homogeneous-phase systems of

polyethylene oxides and rhodium compounds are announced •

• Through-space molecular interaction of f-CH2CH20-1n alters the

electron density at Rh(I), and this effect promotes a representative
oxidative addition to Rh(I) five- to ten-fold •

• Intramolecu1ar insertion of Rh(I) into sp3 carbon-hydrogen bonds

of the polyether Ph2P(CH2CH20)10C16H33 is supported by lH_, l3C{lH}-
and 3lp{lH}_NMR spectral data •

• Intermolecular insertion of Rh(I) into an sp3 carbon-hydrogen

bond occurs at room temperature, atmospheric pressure, and in the
dark. Products of subsequent hydrogen transfer from the insertion
products [of Rh(I) with both Ph2P(CH2CH20hoC16H33 and MeO(CH2CH20ho-
C16H33] are reported.


Lehigh University is thanked for hospitality, which has been

extended during a period as a Visiting Research Scientist.


1. R. J. Daroda, J. R. Blackborow, and G. Wilkinson, J. Chern. Soc. 2

Chern. Commun. , 1098 (1980).
2. ~. C. Baird, J. T. Mague, J. A. Osborn, and G. Wilkinson,
J. Chern. Soc. 2 A, 1347 (1967).
3. T. Nishiguchi and K. Fukuzumi, J. Am. Chern. Soc. , 96:1893 (1974).
4. J. A. Osborn, F. H. Jardine, J. F. Young, and G. Wilkinson,
J. Chern. Soc. 2 A, 1711 (1966).
5. R. D. Schrock and J. A. Osborn, J. Am. Chern. Soc. , 98:2134
6. N. W. Alcock, J. M. Brown, and J. C. Jeffery, J. Chern. Soc.,
Dalton Trans., 888 (1977).
7. N. W. Alcock, J. M. Brown, and J. C. Jeffery, J. Chern. Soc.,
Chern. Commun., 829 (1974).
8. N. W. Alcock, J. M. Brown, and J. C. Jeffery, J. Chern. Soc.,
Dalton Trans., 583 (1976).
9. For specific examples see references 4, 5, and J. A. Tiethof,
J. L. Peterson, and D. W. Meek, Inorg. Chern., 15:1365 (1976).
10. D. W. Watts, Pure App1. Chern., 51:1713 (1979).
11. A. S. Hussey and Y. Takeuchi, J. Am. Chern. Soc., 91:672 (1969).
12. A. S. Hussey and Y. Takeuchi, J. Org. Chern., 35:643 (1970).
13. F. H. Jardine, J. A. Osborn, and G. Wilkinson, J. Chern. Soc., A,
304 M.l.DEEM

1574 (1967).
14. W. Voe1ter and C. Djerassi, Chem. Ber., 101:58 (1968).
15. L. Horner, H. Blithe, and H. Siege1y Tetrahedron Lett., 4023
16. w. Strohmeier and E. Hitze1, J. Organomet. Chem., 87:257 (1975).
17. J. P. Cand1in and A. R. Oldham, Discuss. Faraday Soc., (46):60
18. A. Levi, G. Modena, and G. Scorrano, J. Chem. Soc., Chem.
Commun., 6 (1975).
19. W. S. Knowles, M. J. Sabacky, B. D. Vineyard, and D. J. Wein-
kauff, J. Am. Chem. Soc., 97:2567 (1975).
20. M. D. Fryzuk and P. Bosnich, J. Am. Chem. Soc., 99:6262 (1977).
21. D. G. Constable, C. R. Langrick, B. Shabanzadeh, and B. L. Shaw,
Inorg. Chim. Acta, 65:L151 (1982).
22. Y. Dror and J. Manas sen , J. Mol. Cata1., 2:219 (1977).
23. A. F. Borowski, D. J. Cole-Hamilton, and G. Wilkinson, Nouv. J.
Chim., 2:137 (1978).
24. F. Wuesthoff, E. F. von Pechmann, D. Behrens, and R. Goetz,
German Patent 2,627,354 (Dec.' 23, 1976).
25. R. Bar, Y. Sasson, and J. Blum, J. Mol. Cata1., 16:175 (1982).
26. R. G. Nuzzo, D. Feit1er, and G. M. Whitesides, J. Am. Chem. Soc.,
101:3683 (1979).
27. R. G. Nuzzo, S. L. Haynie, M. E. Wilson, and G. M. Whitesides,
J. Org. Chem., 46:2861 (1981).
28. Y. Dror and J. Manassen, Proc. 7th Int~rn. Congr. Cata1., Tokyo,
July 1-2, 1980, Abstr. B14.
29. S. Franks, F. R. Hartley, and J. R. Chipperfie1d, Inorg. Chem.,
20:3238 (1981).
30. M. L. Deem, unpublished experimental results/submitted for pub-
31. Anal. Ca1cd for C37H76011'3H20: C, 59.17; H, 11.01. Found: C,
59.43; H, 11.18.
32. Anal. Cal cd for C48H83010P,: C, 67.73; H, 9.83; P, 3.64. Found:
C, 67.91; H, 9.83; P, 4.03.
33. J. P. Collman, Acc. Chem. Res., 1:136 (1968).
34. F. H. Jardine, Prog.Inorg. Chem., 28:144 (1981).
35. J. Chatt and B. L. Shaw,Chem. Ind. (London), 931 (1960).
36. M. A. Bennett and P. A. Longstaff, Chem. Ind. (London), 846
37. A. J. Birch and K. A. M. Walker, J. Chem. Soc., C, 1894 (1966).
38. A. Emery, A. C. Oeh1sch1ager, and A. M. Unrau, Tetrahedron Lett.,
4401 (1970).
39. A. Yamada, T. Fukuda, and M. Yamagita, Bull. Chem. Soc. Jpn.,
48:353 (1975).
40. R. Uson, L. A. Oro, R. Sariego, and M. A. Esterve1as, J. Organ-
omet. Chem., 214:399 (1981).
41. T. Nishiguchi, K. Tachi, and K. Fukuzumi, J. Am. Chem. Soc.,
94:8916 (1972).
42. H. Imai, T. Nishiguchi, and K. Fukuzumi, J. Org. Chem., 39:1622

43. D. Beaupere, L. Nadjo, R. Uzan, and P. Bauer, J. Mol. Cata1.,

18:73 (1983).
44. R. Spog1iarich, A. Tencich, J. Kaspar, and M. Graziani, J. Organ-
omet. Chem., 240:453 (1982).
45. V. Z. Sharf, L. K. Freid1in, and V. N. Krutii, Izv. Akad. Nauk
SSSR, Sere Khim., 2264 (1973); Chem. Abstr., 80:59533f (1974).
46. R. Spog1iarich, G. Zassinovich, G. Mestroni, and M. Graziani,
J. Organomet. Chem., 198:81 (1980).
47. D. Beaupere, P. Bauer, L. Nadjo, and R. Uzan, J. Organomet.
Chem., 238:C12 (1982).
48. R. Grigg, T. R. B. Mitchell, and N. Tongpenyai, Synthesis, 442
(1981) •
49. C. Masters, A. A. Kiffen, and J. P. Visser, J. Am. Chem. Soc.,
98:1357 (1976).
50. O. R. Hughes and D. A. Young, J. Am. Chem. Soc., 103:6636 (1981).
51. P. Meakin, J. P. Jesson, and C. A. Tolman, J. Am. Chem. Soc.,
94:3240 (1972).
52. C. A. Tolman, P. Z. Meakin, D. L. Lindner, and J. P. Jesson,
J. Am. Chem. Soc., 96:2762 (1974).
53. These are in the range of 90 - 70 ppm [54,55].
54. P. E. Garrou, J. L. S. Curtis, and G. E. Hartwell, Inorg. Chem.,
15:3094 (1976).
55. C. Crocker, R. J. Errington, A. Markham, C. J. Moulton, K. J.
Odell, and B. L. Shaw, J. Am. Chem. Soc., 102:4373 (1980).
56. P. Meakin and J. P. Jesson, J. Am. Chem. Soc., 96:5751 (1974).
57. B. E. Mann, C. Masters, and B• .L. Shaw, J. Chem. Soc., Dalton
Trans., 704 (1972).
58. D. Milstein, Organometa11ics, 1:1549 (1982).
59. J. A. Tiethof, J. L. Peterson, and D. W. Meek, Inorg. Chem.,
15:1365 (1976).
60. P. W. Clark, P. Hanisch, and A. J. Jones, Inorg. Chem., 18:2067
(1979) •
61. C. Crocker, R. J. Errington, W. S. McDonald, K. J. Odell, and
B. L. Shaw, J. Chem. Soc., Chem. Commun., 498 (1979).
62. P. w. Clark, J. Organomet. Chem., 110:C13 (1976).
63. W. D. Jones, II, cited in J. Haggin, Chem. Eng. News, 61(7):9
(1983) •
64. C. A. Tolman, S. D. Itte1, A. D. English, and J. P. Jesson,
J. Am. Chem. Soc., 101:1742 (1979).
65. J. w. Suggs, J. Am. Chem. Soc., 100:640 (1978).
66. D. Milstein, J. Am. Chem. Soc., 104:5227 (1982).

Jack D. Chase

524 Maplehurst Ave.

Oakville, Ontario
Canada L6L 4Zl


The phasedown of lead in automobile gasoline since the late

1970's has created a need for high octane unleaded gasoline compo-
nents. In response to this need, processes have been developed for
alkyl ethers that are formed by the addition of iso-olefins (more
correctly, tertiary olefins), such as isobutylene and isoamylenes,
to methanol. In fact, the reaction of the former to make methylter-
tiarybutyl ether (MTBE) constitutes the fastest growing use for meth-
anol in the USA and Europe [1-3]. MTBE was first made commercially
in Europe by ANIC using Snamprogetti technology in 1973 and later in
1976 [4,5]. The first plant in the USA was built in 1979. Now there
are at least 14 MTBE plants in the USA (all but 2 use the relatively
concentrated steam cracker feedstock). There are presently no plants
producing tertiaryamylmethyl ether (TAME) from isoamylene and meth-
anol, but this is likely to occur in the future [6].

The constraint on alkyl ether production is the availability

of tertiary olefins. Consequently, the potential for MTBE and TAME
amounts to 3.2% [7] and 2.0%, respectively, of unleaded gasoline
production in the USA. Nonetheless, these rather small percentages
are still large numbers with US unleaded gasoline consumption being
about 3 million barrels/day.

Although demand for gasoline is presently declining due to the

increased popularity of smaller fuel-efficient automobiles, it is
projected that the demand for high octane unleaded gasoline will in-
crease partly as a replacement for metallic antiknock compounds for
environmental reasons and partly due to the desirability to improve
engine efficiency by increasing compression ratio. These ethers,

308 J. D. CHASE

when blended into gasoline at the 10% level, increase research and
motor octane by 2-3 octane numbers.

There are many combinations of tertiary olefins and primary

alcohols that produce alkyl ethers with desirable octane qualities,
but this paper is limited to the two that are economically viable to
produce at the present, MTBE and TAME. Some reaction details includ-
ing kinetics of these catalytic reactions, economic considerations,
and some aspects of the physical properties of these ethers are cov-
ered in this paper.



CH3-C-CH3 + CH30H • • (1)

Isobutylene Methanol Methyltertiarybutyl ether (MTBE)

Catalyst CH3
CH3-CH2=C-CH3 + CH30H • • CH3-CH2-C-O-CH3 (2)

Isoamylene Methanol Tertiaryamylmethyl ether (TAME)

The above reactions occur below 180°C in the liquid phase and
are generally carried out at about the minimum pressure to assure
liquid phase at reaction temperature. There are three isomeric iso-
amylenes: only the 2-methyl butenes react, i.e. 2-methyl-l-butene and
2-methyl-2-butene, while the 3-methyl-l-butene does not react.
The addition reactions with alcohol are very selective and only occur
with the tertiary olefins. Consequently, when the C4 fraction of an
olefin stream from an industrial installation such as an ethylene
plant is passed through an esterification reactor, the other olefins
pass through the process unreacted. Incidentally, this situation has
lead to the proposed scheme [8] for separating isobutylene from other
C4s, whose boiling points are so close as to make distillation sep-
aration difficult. In such a case, the isobutylene is first ester-
ified, then separated from other C4s, and finally decomposed back to
isobutylene and methanol.


Published Snamprogetti work [9] has shown the kinetics to be


first order in isobutylene and zero order in methanol when more than
stoichiometric amounts of methanol are present. This was similarly
reported by work at Englehard [10]. This was confirmed by Gulf
Canada studies where results are plotted in Figure 1. The etheri-
fication of isoamylenes have similar type kinetics with respect to
the orders of the various reactants as reported by Snamprogetti [9].

The ratio of initial etherification reaction rates of isobutyl-

ene and isoamylene is 1.85 from the literature [9]. At 70°C the
initial reaction rate has been found to be 0.012 g moles isoamylene/
hr/g dry catalyst [11]. Consequently, using the above 1.85 ratio
of reaction rates, the initial reaction rate for isobutylene is
0.022 g isobutylene/hr/g dry catalyst. These rates apply for a mix-
ture of 20 wt % tertiary olefin in olefin mixture feedstocks, and
stoichiometric methanol/t-olefin ratio, and the catalyst used in
Reference 9. The effect of temperature on the normalized etherifi-
cation reaction rate of 100% isobutylene hydrocarbon feed for a meth-
anol/isobutylene feed ratio of 2 from Gulf Canada work at high space
velocity is shown in Table 1.

Published results [9] for both etherification of isobutylene

and isoamylene suggest that an ionic mechanism is most probable and
olefin protonation can be seen as the rate determining step. Also,
since bond isomerization occurs as well as etherification of isoamyl-
enes, this implies the existence of a carbonium ion as a common in-

Both the kinetics and the chemical thermodynamics (equilibrium)

are more favorable for ether formation with MTBE than TAME. (Actu-
ally a more appropriate designation of TAME would be MTAE for meth-
yltertiaryamyl ether, but a more pronouncable acronym was used (TAME)
for promoting the Gulf process technology for this ether.) As an
indication of the reversibility of the isoamylene etherification
reaction or the effect of conversion on forward reaction rate, eval-
uation of the reactor design equation below shows required reactor
packed with catalyst (V) is twice as large for conversion X = 0.47
than for X = 0.36, where r is the reaction rate obtained from dif-
ferential data and F is the moles of iso-olefin per hr in the feed.

Reactor volume (V) oc Catal. wt. (W) = FIxo dX/r (3)

The class of catalyst used for much of the etherification work

published in the scientific [9] and patent literature [8] consists
of ion exchange resins in the acid form comprised of styrene-divinyl-
benzene and containing sulfonic acid groups [7]. Amberlyst 15 pro-
duced by Rohm and Haas and used in the Snamprogetti study [9], is
typical of the catalyst used in kinetics reported here. It should
not be assumed though that this is used in the Gulf MTBE and TAME
310 J. D. CHASE

~ 1.2
Q 0 ••
C 0.4
o 1 2 3
(CH 3 0H/ iC4)

Figure 1. The dependence of the reaction rate at 180°F on the

CH30H/iC4 molar ratio (~) with the isobutene inlet con-
centration = 1.0 g mole/I.

Mass Transfer Considerations in Reactor

In order to test whether the reaction was mass transfer or dif-

fusion rather than kinetically controlled or limited, a series of
experiments were carried out at Gulf by varying the Reynolds number a
of flow through the reactor while maintaining reciprocal space veloc-
ity (W/F = weight of catalyst/flow rate of isobutylene) constant and
determining the effect on conversion. If the conversion remained
constant this would indicate a negative effect, i.e. mass transfer
does not control the reaction. This is illustrated in Figure 2.
Another indication that chemical kinetics and not mass transfer is
rate limiting is that identical conversions and reaction rates were
obtained for two different catalyst particle sizes (+16, and +48 mesh
to -35 mesh). This is shown in Figure 3, where results at a constant
reciprocal space velocity (W/F) are plotted as conversions vs tem-
perature for the two catalyst sizes. Since a single line· can be
drawn through data for both particle sizes, this adds further proof
that the reaction is not mass transfer controlled. The low conver-
sions evident in Figures 2 and 3 result from high space velocities
being used in these differential reactor type studies.


Gulf Canada was first to develop and promote [12] process tech-
nology for integrating a MTBE plant into a petroleum refinery using
the less concentrated (10 - 18%) ·isobutylene in the C4 stream from
a refinery cat cracker unit (FCCU).. A major aspect of this technol-

aReynolds number = DVp/~, where V = superficial liquid velocity, p =

liquid density, ~ = liquid viscosity, and D = a distance parameter
and here is catalyst particle diameter.

Table 1. Integral Reactor Data for MTBE Synthesis a

Conversion Temperature
Space Ve10cityb of Reaction Rate C
% °c

31.2 13.5 62 143 1.38

44.1 13.5 80 176 1.91
58.9 13.5 89 192 2.69
76.6 13.5 100 213 3.50
79.1 13.5 107 225 3.52

aHydrocarbon feed is 100% iC4. cp (MeOH! iC4) 2.0;

Selectivity: 97-99%
bBased on iC4
cBased on dry catalyst

ogy involved removal of residual methanol in the unreacted C4s, which

went on to further processing alkylation or polymerization units.
Due to the existence of C4s-methano1 azeotropes, some of which were
heretofore undiscovered [13], distillation could not be used for
methanol removal. A schematic flow sheet for the low (86%) conver-
sion Gulf process, showing glycol adsorption to remove methanol in
the unreacted C4 stream to about 10 ppm, is shown in Figure 4. It
is seen that the low conversion process uses a single catalytic re-
actor. The high (96%) conversion option utilizes two reactors.

Most of the unreacted methanol comes off the bottom of the dis-
tillation column leaving from 1 to 5% methanol with the MTBE product,
depending on process conversions. This product stream typically also
contains up to 1% diisobuty1ene (a good gasoline component) and
about 0.2% C4 hydrocarbons.

All known developed alkyl ether process technology utilizes

cation exchange resins as catalysts [7]. This reference, SRI Inter-
national [7], reports reaction temperatures to range between 30 and
100°C, and reactor pressures between 7 and 14 atmospheres.

The economics of the high conversion (96%) Gulf Canada process

at a scale of 438 BPSDa MTBE are given in Table 2. Costs are in US $
and are believed to be representative of costs in January 1983.

aBPSD = barrels per stream day. This is distinct from barrels per
(calendar) day, since there is usually a 3-week maintenance shut
down annually.
312 J. D. CHASE
III 25
e 20

Z 15
iii 0 0 ~
0 5

~ 0
0 20 40 60 80

Figure 2. Check for mass transfer limitations at 220°F with a re-

ciprocal space velocity based on isobutene of 0.45.



This section is devoted to TAME technology [12,14-16] and is

larger than the MTBE process technology section, since much more has
already been published on MTBE and is available elsewhere [2,5,7,8,

The reaction kinetics for the methoxylation of isoamylenes is

less favorable than for isobutylene methoxylation necessitating
multi-pass recycle design [16], in contrast to the single pass ap-
proach possible for MTBE production. Typical sources of branched C5
olefins are the FCCU, and thermal cracker by-products in ethylene
production. Concentrations of the reactive isoamylenes in light cat
70 80 90 100 °c
~ 25
e 20

~ 15
ffi 10
0 5

160 180 200 220
Figure 3. Effect of temperature and catalyst particle size (0, 35-48
Mesh; 0, +16 Mesh) on the % conversion to MTBE.




Figure 4. A schematic diagram of the methyltertiarybutyl ether

(MTBE) process.

cracked gasoline (LCCG) depend on severity of cracker operation. It

is typically 2S% of the Cs fraction. Products in the gasoline boil-
ing range from thermal crackers, pyrolysis gasoline, are first-stage
hydrotreated and then normally sent to gasoline blending. This
first-stage hydrotreated pyrolysis gasoline (HPG) is a good source
of isoamylenes. Often, the concentration of isoamylenes in HPG is
higher than in light cat cracked gasoline.

Combined C4 and Cs olefin methoxylation in the same reactor has

been pilot planted and patented [21]. However, this approach is less
attractive at present than designs that are utilizing separate reac-
tors for making MTBE and TAME, although another firm has taken the
opposite approach [22]. Nonetheless, there are considerable economic
incentives for making both ethers on the site since off sites and
expertise can be shared [22].

Process Description

A schematic diagram of Gulf Canada's TAME process is given in

Figure S. The generalized diagram is typical although there are a
number of variations or process concepts that will be discussed in
the next section. The olefin feed (LCCG or HPG) is first fraction-
ated to separate the Cs fraction from the C6+ material; the latter
goes to gasoline blending, while the CSs are mixed with methanol and
enter a fixed bed reactor containing cation exchange resin (hydrogen
form) catalyst. The effluent from the liquid phase reactor is split,
and some of this goes to the distillate column where TAME is separated
from unreacted isoamylenes and unreactable Cs hydrocarbons, which are
recycled to the reactor. The fraction of reactor effluent that is

Table 2. MTBE Economics with the Gulf Canada Process; CAT Cracked

1. Basis: 22,000 BPSD FCCU - 438 BPSD MTBE

2. Onsite capital cost of $1.74MM (USA Gulf Coast 1983)
3. Production cost: ¢/US gal MTBE
A. Materials & chemicals (MeOH @ 67¢ gal) 25.91
B. Utilities (steam @ $6.l6/MM BTU) 4.24
C. Labor (1 man/shift) 1.90
Indirect cost (includes depreciation-14 years) 6.50
Production Cost (excluding isobutylene) 38.55

not recycled is either fed directly to the gasoline pool or passed

through an optional methanol removal unit. The latter consists of
a counter-current extraction column whereby the reactor effluent
stream, containing a few percent methanol, enters the bottom of the
extractor. The TAME-containing raffinate stream from the extractor
contains from 7 to 15% TAME, not more than 0.1 wt% methanol, and is
suitable for gasoline blending. The methanol-rich extract goes to
a simple packed stripping column where the methanol is distilled
overhead and recycled to the reactor, while the methanol-lean extract
is recycled to the top of extractor.

The location of the bleed stream to purge off the lighter ends
is at the reactor exit. As shown in Figure 5, a single distillation
column can serve as initial feed depentanizer and as an ether column
for separating TAME and C5 hydrocarbons. This doubling-up of service
provides both operating and capital cost savings, in comparison with
a TAME process with depentanizer and ether column upstre~ and down-
stream of the reactor, respectively. However, this economy may be
bought at the cost of loss of some blending flexibility, since the
ether product stream is more concentrated (greater than 50% TAME) in
the two column case, compared to being less than 5% TAME (remainder
being C6+ hydrocarbons) in the single column case.

Specific Processing Options

Within the framework of the TAME process, there are various

process concepts or options available. It is worth noting that there
is a strong economic incentive to allow as much high octane, rela-
tively cheap, methanol in the gasoline as is consistent with gasoline
product specifications. The TAME process conversions are controlled
by the amount of C5 hydrocarbon recycled, and under present economic
conditions the optimum conversion will be within the range of 40-70%.
The processing options discussed so far have been confined to "add-







Figure 5. A schematic diagram of the tertiaryamylmethyl ether (TAME)


on" rather than an integrated process. The former simply are added
to the olefinic gasoline stream in a refining complex. By slightly
modifying the upstream octane splitter column, which separates the
FCC gasolines into heavy cat cracked gasolines (HCCG) and light cat
cracked gasolines (LCCG), the economics of TAME processing are sig-
nificantly improved.

The modification of the octane-splitter envisaged consists of

changing the cut point such that it operates as a depentanizer and
only the C5 hydrocarbons are taken overhead. This eliminates the
need for a separate depentanizer in the two column process.


Production cost break-downs are given for three process concepts

that are defined as follows:

(a) Concept 2: High conversion (69%) - high recycle, already

depentanized feed without methanol removal unit, - 808 BPSD of TAME.

(b) Concept 3: Low conversion (42%) - low recycle, already

depentanized feed, with methanol removal unit, - 483 BPSD of TAME.

(c) Concept 4: High conversion (69%) - high recycle, full range

LCCG feed, with methanol removal unit, - 808 BPSD of TAME.

The cost break-downs of Concepts 2,3, and 4 are given in Tables

316 J. D. CHASE

3, 4, and 5, respectively. The cost of isoamylene feedstock is not

included, consequently these figures refer to isoamylene upgrading

The assumptions made in the economics calculation are listed

below. (a) and (b) have been used throughout this paper.

(a) Value of 1 US gal of 100 RON unleaded gasoline = $1.05.

(b) Value of 1 octane-bbl RON unleaded of 100 39¢.
(c) The RON of isoamylene = 96 (consequently, 1 gal of isoamyl-
ene is worth $1.01).
(d) TAME (Concept 2) process at 808 BPSD TAME capacity.

These assumptions are felt representative of costs in the USA

in January 1983. In order to make return on investment calculations,
revenue evaluations were determined using incremental revenue above
that for isoamylene, and consequently this is consistent with pro-
duction costs given on this basis in Tables 2, 3, 4, and 5.

In Table 6, a comparison of various aspects of the different

TAME process concepts including return on investment is given in con-
junction with the high conversion MTBE process for cat cracker feed-
stock. It is seen that in 1983, TAME production is economical based
on C5s-only olefin feedstock but marginal/break-even for full range
light cat cracked gasoline, based on the particular input assumptions
used. This contrasts with similar economics done in 1980 on prevail-
ing US octane and methanol value at that time [15]. The most econom-
ic TAME process concept involves low recycle (low conversion).

It is noteworthy that although the MTBE process provides a

better return on investment than TAME processing, .even the former
provides lower returns than during years 1978-1982 (higher gasoline
production). Consequently, the data in Table 6 is provided mainly
for purposes of comparison, and to provide some current figures, and
certainly not to serve as an indicator of future projections. In
the current climate of wide swings in gasoline costs, the short term
economics of TAME processing can fluctuate, but the author believes,
in the long term, TAME will be found as a commercial gasoline com-


Some relevant properties of MTBE and TAME are listed in Table

7. Blending octane of the ethers varies somewhat depending on what
base gasoline into which the ether is blended. The Gulf [12] figures
are average values from measurements made for 10% ethers in various
base gasolines, which compare closely to averages of RON/MON = 113.5/
100.5 for the range of values given elsewhere [22]. Although the
blending octane of TAME is somewhat lower than MTBE, TAME has the

Table 3. TAME Economics - Concept 2 - 808 BPSD TAME

1. High recycle, C5s only feed, no MeOH removal

2. Onsite capital cost = $2.l6MM (USA Gulf Coast 1983)
3. Production cost: e/US gal TAME
A. Materials & chemicals (MeOH @ 67e gal) 34.68
B. Utilities (steam @ $5.00/MM BTU) 9.63
C. Labor (1 man/shift) 1.34
Indirect cost (includes depreciation-14 years) 4.09
Production cost (isoamylene excluded) 49.74

Table 4. TAME Economics - Concept 3 - 483 BPSD TAME

1. Low recycle, C5s only feed, with MeOH removal

2. Onsite capital cost: $0.53MM (USA Gulf Coast 1983)
3. Production cost: e/us gal TAME
A. Materials (MeOH @ $67/gal) 23.20
B. Utilities (steam @ $5.00/MM BTU) 3.70
C. Labor (1 man/shift) 1.80
Indirect cost (includes depreciation-14 years) 1.55
Production cost (isoamylene excluded) 30.25

Table 5. Ether Production Costs & Assumptions a (USA, 1983)

Concept MeOH Cost Production Cost b

Value of Octane-BBL
(e/US gal) (e/US gal Ether)

1 MTBE 67 3ge 38.6

2 TAME 67 3ge 49.7
3 TAME 67 3ge 30.2
4 TAME 67 3ge 56.6

aAssumed value of 1 US gal 100 RON unleaded gasoline = $1.05

bIncludes MeOH and indirect costs, excludes iso-olefin

Table 6. Return on Investment & Details for Various Ether Concepts

Process Concept 2TMffi 3TMffi 4 TAME

Process feed C4s, LCCG or Css C4 s CSs Css LCCG

Process scale BPSD of ether 438 808 483 808
Iso-olefin conversion (%) 96 69 42 69
Volume yield ether olefin 1.27 .873 .S18 .873
Methanol in product stream (wt%) 2.2 4.S 0.37 4.S
ROlb (after tax, first year) 22 S.S l6.S

aHigh (96%) conversion process

bCosts based on USA Gulf Coast in 1983

compensating advantage of lower vapor pressure.

There is considerable interest in developing solubilizers for

blends of arcund 10% methanol in gasoline, since methanol is both
less expensive and has higher octane than gasoline. The problem with
such blends with the polar methanol is that small amounts of water
(about 0.01%) can cause methanol and gasoline to form separate liquid
phases. To some extent, alkyl ethers act as solubilizers for meth-
anol-gasoline blends by increasing the low water tolerance for these
mixtures, particularly at low temperature.

Results of experiments carried out in a cold-room at Gulf at

four temperatures from -lSoF to 70°F (-26 to +2l0C) for 10% TMffi
added to mixtures of 3.7 and 6 vol % methanol in aromatic base gas-
oline are shown in Tables 8 and 9. The water tolerances at -lSoF are
seen to vary between about 300 and 1,000 ppm water. The percentage
increases in water tolerance for these mixtures are shown to be be-
tween 100 and 200% in the important low temperature range. Although
MTBE increases water tolerance more than TMffi in methanol-gasoline
mixtures, the latter was measured more thoroughly due to process
development considerations.


Most of the isobutylene made is produced in the catalytic crack-

ers in oil refineries and is typically upgraded to gasoline by alkyl-
ation or polymerization. The volumetric yields and product octane
(quality) for the various methods of upgrading isobutylene are given
in Table 10. From this table, it is seen that on the basis of 1
volume of olefin (isobutylene), only 0.43 volume of methanol is needed
to make 1.27 volumes of MTBE with a product octane of 110. Whereas,
Table 7. Properties of MTBE and TAME :0
Property MTBE TAME Reference ~
Molecular weight 88.15 102.17 :x>
Density (g/cc at 20°C) .7405 .7703 23 Z
Vapor pressure (rom Hg at 25°C) 245 75 23
Heat of vaporization (cal/g) 76.8 78.0 23 :x>
Solubility of water in ether (wt % at 20°C) 1.3 0.6 23 o
Boiling point (OC) 86.3 23 CJ)
55 o
Wt % ether in MeOH-ether azeotrope 85 50 23 6
Octane RON/MON 118/101 112/99 12 r
Octane RON/MON 105-122/96-105 22 :!!
Refractive index (20°C) 1.3689 1.3885 23 z


Table 8. Water Tolerance (Vol %) of Oxygenated Gasoline Blendsa at

Two Temperatures

Base + 3.7% MeOH 0.030 0.036

Base + 3.7% MeOH + 10% TAME 0.089 0.094
Base + 6% MeOH 0.042 0.052
Base + 6% MeOH + 10% TAME 0.107 0.117

aAromatic base gasoline contained 57% saturates & 43% aromatics

Table 9. Percent Water Tolerance Increase Due to Addition of TAME

to Gasoline Methanol Blendsa,b

% Water Tolerance Increase

% %
MeOH TAME at Various Tem~eratures
-15°F O°F 32°F 70°F

3.7 10 197 161 133 89

6 10 155 125 93 78

aAromatic base gasoline contained 57% saturates & 43% aromatics

~BE increases water tolerance more than TAME

1.2 volumes of isobutane are needed to make 1.78 volumes of alkylate

product of 94 octane. Thus, in comparing methanol to isobutane, the
following ratio is obtained: (1.27(110)/.43)/(1.78(94)/1.2) = 2.3.
That is, 1 barrel of methanol is equivalent to 2.3 barrels isobutane
for upgrading isobutyleneto gasoline. This simple calculation ex-
plains why MTBE production has grown so dramatically during the


Much of the work reported in this paper was carried out between
1974 and 1982 while the author was employed as a Senior Research
Engineer and Project Leader of Oxygasoline Process Development in
the Research and Development Department of Gulf Canada Limited. The
author, who is working as a consultant in early 1983, wishes to thank
Gulf Canada for permission to publish this work.

Table 10. Results of Upgrading Isobutylene

Chemical Equation Product Octane

(R + M)/2 R/M

a) Alkylation:
isobutylene + isobutane + trimethyl pentane 94 95/92
1 1.2 1. 78 volumes
b) Polymerization (Polygas):
isobutylene + n-butene + iso octene 89 98/80
1 1.0 1.62 volumes
isobutylene + methanol + MTBE 110 118/101
1 0.43 1.27 volumes


1. J. C. Davis and P. M. Kohn, Chern. Eng., 91 (May 21, 1979).

2. S. C. Stinson, Chern. Eng. News, 35 (June 25, 1979).
3. Chem. Eng. News, 27 (Jan. 24, 1983).
4. Chem. Week, 39 (Dec. 1, 1976).
5. Hydrocarbon Process., 61(9):177 (1982).
6. J. D. Chase and B. B. Galvez, National Pet. Refiners Assoc.
Meeting, New Orleans, Louisiana, (March 23-25, 1980); Abstr.
No. AM-80-46.
7. A. C. Gaess1er, CEH Product Review (675.1000 C), Chemical Eco-
nomics Handbook, Stanford Research Institute International,
Menlo Park, California (June 1981).
8. J. A. Verdo1, U.S. Patent 3,170,000 (1965); assigned to Sinclair
9. F. Ance110tti, M. M. Mauri, and E. Pescar01lo, J. Cata1., 46:49
(1977) •
10. A. E. Eleazar, R. A. Heck, R. G. Mc Clung, and B. A. Olson, 88th
National AIChE Meeting (June 9, 1980).
11. Gulf Canada, unpublished data.
12. J. D. Chase and H. J. Woods, 176th National ACS Meeting, Div.
Pet. Chern., ACS, Miami Beach, FL, Abstr. No. PETR-61 (1978).
13. J. D. Chase, B. B. Galvez, and B. W. Kennedy, U.S. Patent
4,218,569 (1980); assigned to Gulf Canada, Ltd.
14. J. D. Chase and H. J. Woods, Oil Gas J., 77(15):149 (1979).
15. J. D. Chase and B. B. Galvez, Hydrocarbon Process., 60(3):89
16. Chemische Werke Hu1s Aktiengesel1schaft Presentation, 86th
National AIChE Meeting, Philadelphia (June 1978).

17. Platts' Oi1gram News (March 20, 1979).

18. A. F. Talbot, American Petroleum Institute, 44th Refining Meet-
ing, San Francisco, California (May 14-17, 1979).
19. G. L."Ke11er, J. H. Bandino, and K. L. Boekhaus, National Pet.
Refiners Assoc. Meeting, Houston, Texas (Nov. 1-2, 1979).
20. G. H. Unze1man and G. W. Mecha1ski, National Pet. Refiners
Assoc. Meeting, San Antonio, Texas (March 27, 1979).
21. H. J. Woods, J. D. Chase, and B. B. Galvez, U.S. Patent 4,204,077
(1980); assigned to Gulf Canada, Ltd.
22. B. Torck, A. Convers, and A. Chauve1, (IFP) AIChE Meeting,
Detroit, MI (Aug. 16-19, 1981).
23. T. W. Evans and K. R. Edlund, Ind. Eng. Chem., 28:1186 (1936).


Lujia Liu, Ricardo Garza Tobias, Kenneth McLaughlin,

and Rayford G. Anthony*

Department of Chemical Engineering

Kinetics, Catalysis and Reaction Engineering Laboratory
Texas A&M University
College Station, Texas 77843


A unique process for producing ethylene, propylene and propane

from coal is to gasify the coal to produce synthesis gas, to convert
the synthesis gas to methanol, and to convert methanol to the low
molecular weight hydrocarbons. Of course, any organic material could
be used to produce the synthesis gas for subsequent conversion to
methanol. Koch [1] concludes that for the production of C2 - C3 ole-
fins from a coal-derived synthesis gas, conversion of methanol is the
most efficient process in terms of yield, carbon usage and thermal

This paper presents a review of the literature and new data,

which were obtained by using a continuous fluidized-bed reactor, for
conversion of methanol to low molecular weight olefins. The exper-
iments with the fluidized-bed reactor were designed to study the
effect on the product distribution and yield for methanol conversion
to low molecular weight hydrocarbons of the following variables:
(1) dilution of the feed and catalyst residence times; (2) type of
feed diluent; (3) temperature and weight hourly space velocity
(WHSV); and (4) the amount of coke and the rate of coking.


Dehydration of methanol usually results in the production of

dimethy1ether and water. However, hydrocarbons were detected in the
reaction products in studies of methanol dehydration over concentrated

324 L. LlU ET AL.

sulfuric acid [2], activated alumina [3], cation exchanged faujasites

[4,5], H-mordenite [6], phosphorous pentoxide [7], and polyphosphoric
acid [7]. Chang and Silvestri [8] report on the use of H-ZSM-5 for
converting methanol, other alcohols, dimethylether, aldehydes, ace-
tone, n-propylacetate, n-butylformate and methanethiol to hydrocar-
bons. Product distributions for the reactants, methanol and dimeth-
ylether, are reported as a function of space velocity and temperature.
At low methanol or dimethylether conversions the products were ole-
fins, and at 100% conversion the products were composed primarily
of C6+ paraffins (and olefins) and aromatics.

High yields, 70 to 90% of ethylene, propylene and propane, at

100% conversion of methanol were first reported by Lin, Chao, and
Anthony [9]. Union Carbide's small pore zeolite, AW500, ion ex-
changed with NH4 and rare earth chlorides was the catalyst. Chang,
Lang, and Silvestri [10] were issued a patent that claims discovery
of chabazite, the principle component of Union Carbide's AW500, er-
ionite, zeoli;e T, zeolite ZK-5 and other aluminosilicate zeolites
with the capability of converting methanol to low molecular weight
olefins. Givens et al. [11] claim the use of erionite, offretite,
zeolite T and ZSM-34, a unique crystalline alumino silicate zeolite
of the erionite-offretite family, for selectively converting metha-
nol and dimethylether to low molecular weight olefins. They claim
the use of steam as a diluent in the methanol feed to enhance the
yield of ethylene. Singh, Lin, and Anthony [12] and Anthony and
Singh [13] also report the use of steam as a diluent for enhancing
the yields of ethylene from methanol with ion exchanged AW500.

The use of carbon disulfide at concentrations less than 3000

ppm to reduce the rate of catalyst decay due to coking was also re-
ported by Singh, Lin, and Anthony [12]. At CS2 concentrations of
6000 ppm the reaction products were primarily CO and H2.

Cobb et al. [14] and Tipnis [15] reported on the use of Zeolon
500, a Norton product composed of erionite and chabazite, for meth-
anol conversion to low molecular weight olefins. The extent of
lanthanum ion exchange had little effect on the product distributions.
The catalyst activity changed with time on stream with paraffins
being the first product, olefins were present at intermediate times
and dimethyl ether was the major product toward the end of the reac-
tion period. The aged catalyst sample contained approximately 10%
coke. Cobb et al. [14] report substantially higher methane and C4
concentrations than reported by others as shown in Table 1. This
may be due to the existence of significant exotherms in the tubular
reactor that was used.

Inui et al. [16] report the optimum crystallization time in

preparation ofZSM-34 is 25 days for maximum yield of ethylene from
methanol. They also found that ZSM-34 doped with thorium gave high
yields of ethylene and propylene (see Table 1).

Table la. Product Distributions with Small Pore Zeolites 6

H-erionitea erionite In
Catalyst H-Chabazitea H-erionite a Zeolite Ta -I
Dealuminized orea ::I:
Temp., oK 811 643 811 643 643 643 643 -I
0.5 2.6 0
WHSV 1.3 1.3 1.3 r
Conversion 100 3 100 5 80 27.4 3.3 0
Hydrocarbon ~
Distribution, 0
wt% m
Methane 3.3 10 11.0 6 2 13 2.1 c
Ethane 4.4 14 5.0 5 1 0.2 0 »
Ethylene 25.4 32 26.7 44 46 22 40.5 0
Propylene 21.2 31 18.8 33 26 31.9 22.1 r
Propane 33.3 2 27.0 1 2 8.7 0
Butane 1 1 4 5.7 22.9 Z

{ 10.4} { 9.6}
Butene 8 10 10.3 7.4
C! 2.0 9 1.9 2 9 8.2 4.6

aReference 10, Chang et al.

Table lb. Product Distributions with Small Pore Zeolites w
ZSM-34 Ion exchanged
Catalyst ZSM-34 a ZK5 c Zeo1on 500 e
with Thb AW500 d

Temp., oK 644 644 673 700 680g 690 h 673 673

WHSV 2.0 3.6 f 0.5 0.10 0.15 0.197 0.943
Conversion 88.2 72.2 100 100 100 85 91 90
Methane 2.1 1.6 7.3 3.2 5.9 6.4 18.2 14.8
Ethane 0.4 1.1 0 3.8 7.6 1.0 0.7
Ethylene 42.5 59.7 39.6 21.4 33.3 38.8 31.1 25.1
Propylene 26.1 23.6 36.9 13.5 32.1 26.0 21.2 19.3
Propane 1.8 5.2 7.8 31.8 21.8 21.2 5.3 3.2
Butane 3.8 1.3 2.3 7.1 17 .9 23.7
Butene 6.7 5.8 6.1 2.8 3.6 13.3
c5 16.5 1.7 7.5 0.3

aReference 11, Givens et a1.

bReference 16, lnui et a1., Feed was 12% by mol CH30H/88% N2, WHSV is total for combined feed.
cReference 10, Chang et a1.
dReferences 9 and 12, Anthony et. a1. (Approximately 10% DME was in the product; CO and CO 2 ~ 4%)
eReferences 14 and 15, Cobb et a1. r
f1.2 = WHSV for CH30H, 2.4 = WHSV for H20, 3.6 = total WHSV. For the CH30H feed the product dis- I

tribution was 36.9% DME, 40.3% H20 and 22.4% hydrocarbons. For the methanol/water feed the C
product distribution exclusive of the feed was 3.9% DME, 54.7% H20 and 41.4% HC. -I
gTubu1ar reactor. »
hBerty reactor (internal recycle reactor).

Ceckiewicz [17,18] studied the decationated form of Zeolite T

(erionite - offretite) and found the rate of coking to increase with
an increase in decationization of Zeolite T. For a Zeolite T sample,
88% decationized and 6% dea1uminized, pulse reactor experiments of
450°C yielded methanol conversion to hydrocarbons of 34%. The unre-
acted methanol was 1% and methanol converted to carbonaceous depos-
its was 65%. The gas distribution was 38% (by wt) C2H4, 39% C3H6,
4.8% C3HS, 1.1% C4H10, 5.7% C4H8, 3.9% C5H10, 1% (CH3)20, and 6.5%
CH30H. The ethylene and propylene content of the hydrocarbon dis-
tribution on a coke free base was 41% and 42.2%, respectively.

Each of the above studies utilize the shape selectivity of the

small pore zeolites. Typical product distributions are reported in
Table 1. Hydrocarbon molecules produced inside the zeolite cage that
are larger than the pore opening result in an accumulation of coke
on the catalyst. This results in a decrease in activity, which can
be restored by burning the coke with air. Of course, overheating of
the catalyst will cause a loss of crystal structure with subsequent
loss of catalyst activity. Increasing the silica-alumina ratio by
steaming or in the initial preparation of the catalyst, as in the
case of ZSM-34, appears to reduce the rate of coking. The enhanced
selectivity of ethylene when cofeeding steam may be due to a decrease
in the rate of coking. However, previous studies do not adequately
address whether the steam diluent effect is due to a decrease in the
partial pressure of the reactants, to a thermodynamic effect [19,20]
or whether it actua11~ blocks active sites for production of larger

Marosi et a1. [21] claim the use of H-ZSM-5 for methanol and/or
dimethy1ether diluted with 50% (by wt) of steam for selectively pro-
ducing ethylene. Experiments with dimethy1ether diluted with water,
carbon dioxide, nitrogen or helium yielded a gas composed of 74%,
54%, 39%, and 37% (mo1e%) ethylene, respectively. Propylene content
was 20%, 27%, 33%, and 37% (mo1e%), respectively. The butenes, bu-
tanes, methane, and propane were less than 1%. Presumably, the re-
maining gas would be methanol or unreacted dimethy1ether. This ex-
ample indicates that the action of water is more than just a diluent.
For dimethy1ether the affect may be a simple mass action effect ac-
cording to the reaction


Formation of propylene and higher olefins has been postulated to

occur by reaction of dimethy1ether as follows [16]:

CH3c+H2 + (CH3)20 + CH3C+HCH3 + CH30H (2)

CH3C+HCH3 + C3H6 + ~(a). (3)

A reduction in dimethy1ether partial pressure by reaction (1) and a

328 L. lIU ET AL.

potential inhibiting effect of methanol in reaction (2) would tend

to enhance the yields of ethylene. Water has also been reported by
Wolthuizen et al. [20] to inhibit the polymerization of the ethylene
on H-ZSM-5.

Wunder et al. [22] claim the use of the molecular sieve, l3X,
containing manganese and/or magnesium for conversion of methanol
and/or dimethylether to obtain high yields of ethylene. A hydrocar-
distribution (Ex. 2) of 46.9% by wt ethylene, 29.2% propylene, 12.3%
methane, 5.3% butene, 3.7% ethane, 0.5% propane, and 2.1% butane was
obtained for a methanol feed, a reaction temperature of 400°C, and
reaction pressure of 1 bar. The methanol conversion was 90.1%, but
the conversion to hydrocarbons was only 12.5% with 47.2% conversion
to dimethylether.


By using the procedure presented by Smith and Missen [23], the

following set of independent reactions are readily obtained.
l\GO l\Ho
kcal/mol at 700 0 K

nCH30H ~ (CH2)n + nH20; n=2,3,4, ••• n=2 -27.48 -5.51 (4)

n=3 -44.64 -22.18
n=4 -57.74 -35.83

2CH30H ~ (CH3)20 + H2O -2.18 -4.74 (5)

CH30H ~ CO + 2H2 -16.69 -24.47 (6)

CO + H20 ~ C02 + H2 -3.05 -9.05 (7)

nCH30H + H2 ~ CnH2n+2 + nH20 n=l -28.14 -28.23 (8)

n=2 -39.86 -40.21
n=3 -52.31 -5Z.97
n=4 -66.13 -67.0

(CH2)n ~ nC + nHZ n=2 -22.68 -10.15 (9)

n=3 -30.60 -1.31
n=4 -42.58 4.51

2CO ~ CO 2 + C -11.44 -41.35 (10)

Even though the above reactions represents the stoichiometry,

kinetically the following reactions plus others not listed may be

kca1/mo1 at 700 0 K

( CH3)2 0 + C2 H4 + H2 O -25.3 -0.77 (11)

2(CH3)20 + (C 4H8 ) + 2H20 -70.96 -26.35 (12)

2(CH3)20 + C3H6 + CH30H + H2O -40.28 -12.7 (13)

(CH3)20 + CH4 + CO + H2 -42.65 0.98 (14)

CH30H + CH20 + H2 1.25 21.Z7 (15)

(CHZ)j + (CHZ)n + CnH Zn+Z + CjHZj_Z n=j=2 10.04 9.49 (16)

n=j=3 12.09 1Z.3Z
n=j=4 10.83 9.86

Reactions (9) and (10), by inference, represent the coke accu-

mulation on the catalyst as carbon. The greater probability, how-
ever, is for the olefins to undergo hydrogen (or hydride) transfer,
which produces the paraffins as in reaction (16). The diene specie
then dimerizes to produce cyclic compounds that undergo additional
hydride transfer with low molecular weight olefins producing aromat-
ics and paraffins. Even though reaction (16) is unfavorable, the
formation of aromatics and paraffins from olefins is thermodynami-
cally very favorable. Because of the small pore diameter the aromat-
ic, cyclic, and long chain species are unable to diffuse from the
catalyst and hence accumulate therein forming coke.


Very little has been published on the kinetics of methanol con-

version to low molecular weight olefins. Cobb et a1. [14] for ion
exchanged Zeo1on 500 present a second order model of the form


This expression was combined with the material balance for a plug
flow reactor, integrated and rearranged to yield

akCM8 (18)


k' = kCM
PMw- ]-1
LHSV liquid hourly space velocity
330 l. lIU ET Al.

= molar density of liquid feed

= mole fraction of methanol in the feed. Cobb et al.
used only pure methanol, therefore, YM = 1.
= gas law constant
conversion of methanol
catalyst time on stream
= molecular weight of the feed
= defined by Eq. (17).

Data at 673°K and liquid hourly space velocities of 0.05 to 0.943

hr- l were used to develop the correlation. At zero time on stream,
second order kinetics were observed.

Anthony et al. [24] presented a rate equation of the form


for methanol conversion over AW500 ion exchanged with NH1; and rare
earth chlorides. Data obtained by using a Berty internal recycle
reactor were used to develop equation (19). The activity factor (l
had the functional form,

(l = -=-l-+--'-bt"";:/;"'(:--LH""""S-V"""")
By combining the material balance for the well-mixed reactor with
Equation (19) and the definition of (l and rearranging, the result is
k O p2
1 M , (20)
(1 + bt/LHSV + ki PM)3


The values of ki, ki,

and b could not be correlated with Arrhen-
ius functions [24]. The above rate equation, however, fit the data
for the temperatures of 567, 623, and 69l o K. Hougen and Watson [25]
present a similar function for propane cracking.

The rate equations presented by Cobb et al. and Anthony et al.

are significantly different. Equation (17) indicates that a signif-
icant increase in the rate over a fresh catalyst could be achieved
by increasing the reaction pressure. Whereas, for equation (19), a
reduction in the rate with increased reaction pressure occurs. As
the partial pressure of methanol increases the rate of disappearance
of methanol decreases. Secondly, equation (17) indicates deactiva-
tion can be treated as separable, whereas, equation (19) shows non-

separable kinetics and deactivation. The cubic term in equation

(19) is somewhat surprising in that it suggests the participation of
three active sites in the reaction. However, in methanol synthesis,
the rate equation published by ·Natta [26) has the denominator cubed.
The failure of the rate constants in equations (17) or (19) to fit
Arrhenius functions illustrates that considerably more work is needed
to develop rate equations for methanol conversion to hydrocarbons.

Chen and Regan [27) postulate the occurence of autocatalytic

reactions when using ZSM-5. The kinetic sequence is represented as

A~C (21)

A+C~C (autocatalytic step) (22)

C~D (23)

Where A = alcohol and ether, C olefins, and D paraffins, aro-

matics, and coke.

The reaction (21) is slow relative to reaction (22). This fact

was used to explain the relative high propylene concentrations that
were observed. Chang [28) used a slight variation of the reaction
sequence presented by Chen and Regan to predict the product distri-
bution of olefins, aromatics, and paraffins. Anthony [29) gives a
correction to Chang's equations.

Mihail et al. [30] have developed a kinetic sequence of 53 re-

actions and determined the rate constants for methanol conversion to
gasoline over ZSM-5. The diradical :CH2 and carbenium ions are used
in the mechanism. The first steps in the sequence may be applicable
to small pore zeolites. The reactions are

CH30H ~ co + 2H2 (24)

2CH30H ~ CH30CH3 + H20 (25)

(CH3)20 ~ 2(: CH2) + H20 (26)

: CH2 + CH30H ~ C2H4 + H20 (27)

: CH2 + (CH3)20 ~ C2H4 + CH30H (28)

: CH 2 + ( CH3)2 0 ~ C3 H6 + H2 0 (29)

Higher olefins are formed by

: CH 2 + CnH2n ~ Cn+l H2n+2' where n = 2,3,4,5, and 6. (30)

Reactions (24) through (26) were weakly dependent on reaction temper-

332 L. LlU ET AL.

ature with activation energies of 41,850 to 9,207 kJ/mole. The rate

constants for reactions (27) through (30) were not functions of
temperature, and reaction (28) was twice as fast as reactions (27)
or (29). The rate constant for formation of propylene by reaction
(30) was twice as fast as (28) and four times greater than reaction
(29), which accounts for the high propylene concentrations at low
conversion over ZSM-5. Formation of dienes, aromatics, and paraffins
is by reaction of carbenium ions with olefins by condensation and by
alkylation of aromatics with methanol. In the small pore zeolites
these reactions result in the formation of coke and paraffins.


No mechanistic work has been reported when using small pore

zeolites. However, the chemistry is likely to be the same on all
zeolites with the major differences caused by pore size and acidity.
Therefore, mechanisms published for the initial reactions of metha-
nol decomposition over H-ZSM-5 will be discussed.

Several mechanisms have been presented for methanol and dimeth-

ylether (DME), which lack S-hydrogens, to form olefins. These mech-
anisms are summarized as follows:

1. Initial a-elimination followed by polymerization was proposed

by Venuto and Landis [4]

H-CH2-0H + H20 + : CH2

n[:CH2] + (CH2)n (this step is discounted as not being very

likely by Chang and Silvestri [8]).

2. Swabb and Gates [6] postulated a concerted a-elimination

involving Bronsted acidic and basic lattice sites

0+ H e:+
Basic I Bronsted
O---H---CH 2---O---H---O
Site I I Acid Site

3. Chang and Silvestri [8] support a pathway involving a car-

benoid specie with subsequent reaction with methanol and/or dimeth-
ylether via sp3 C-H insertion. This was represented as

+ ~~2 + CH3CH 20R' + HOR
.. .. ,
H - - - - ·CH20R

CH3CH20R' ~ C2H4 + HOR'


4. Van Hooff [19), Wolthuizen et al. [20), and Dejaifue et al.

[31) postulate formation of ethylene and propylene by primary reac-
tions that proceed via trialkyl o~onium ion intermediates [32).
Subsequent reactions of olefins and carbenium ions occur to form the
higher molecular weight compounds and paraffins.

Perot et al. [33) conducted carbon-13 tracer experiments of di-

methyl ether conversion to hydrocarbons over ZSM-5 and silica-alumina
and obtained data to support ethylene formation through rearrangement
of the trimethyloxonium ion (TMO). The TMO ion was proposed to be
formed between a protonated molecule of methanol or dimethylether
and dimethyl ether according to Kaeding and Butter [34). Ethylene is
a less reactive specie than propylene in these reactions. Anderson
et al. [35) also predict a lower reactivity of ethylene than propyl-
ene on ZSM-5.

5. Ono [36) proposes that the active centers are Bronsted sites
and writes
- 0+
CH30H + HOZ -+ Z-Oo - CH3 + H20
0- 0+ -[H...)+
Z-O - CH3 + CH30H -+ Z-O CH3' : CH2-0H

CH3CH20H(HOZ) -+ C2R4 + H20

The key difference in this mechanism and that proposed by Chang

and Silvestri [8) is that basic sites are needed in the latter mech-

6. Langner [37) studied the methanol reaction over NaH-Y ammoni-

um exchanged 6%, 28%, and 70% Hydrogen-T zeolite, H-ZSM-5, and an
exchanged K-L (Linde type SKH5), and observed the following: (1) In-
crease in temperature favors production of linear molecules; (2) An
increase in acidity increased the hydrogen transfer reaction thereby
increasing the paraffin/olefin ratio; (3) The paraffin/olefin ratio
decreased with increasing temperature; (4) Coke formation and cata-
lyst deactivation was related to pore blockage and formation of meth-
ane at the higher temperatures and to strongly adsorbed species at
the lower temperatures; and (5) Branched hydrocarbons and transfer
reactions involving branched hydrocarbons were more prevalent for the
larger pore zeolites. He then proposed a mechanism that involved a
formaldehyde intermediate on the surface. Lin, Chao, and Anthony [9)
have also proposed formaldehyde species on the catalyst surface, but
the two mechanisms differ in that Langner's [37) proposed mechanism
requires an impurity.

Much is yet to be done to establish the detailed mechanism of

methanol conversion to hydrocarbons.
334 L. LlU ET AL.


The catalyst, AW-500 purchased from Union Carbide, was ion ex-
changed by using a solution of ammonium and rare earth chlorides.
The procedure was similar to that reported by Singh [38] and is de-
scribed in detail by Garza [39]. The exception was that a heated
ion exchange column was used. This enabled preparation of 5 kg of
catalyst for use in the fluid bed unit.

A schematic diagram of the fluid bed unit is shown in Figure 1.

The reactor was a 2 inch diameter (i.d.) No. 316 stainless steel
pipe, and was heated by an electrical resistance furnace. A 40 micron
porous metal distributor was installed close to the bottom. Catalyst
was fed by gravity flow from the feeder through a 1/2" tube into the
reactor. The tube extended into the reactor 16". A vibrator was
used to keep the catalyst in continuous flow, and a clamp was ~sed
to control the flow rate. Methanol was fed by using a metering pump.
Nitrogen was mixed into the line, if necessary, before the feedstock
passed through the preheater, which was a two inch diameter and 9 1/2
inch length stainless steel tube packed with porcelain rings and
heated by a thermal tape. The vaporized gaseous feedstock then passed
through the distributor, entered the reactor, and the conversion oc-
curred. The catalyst particles were fluidized by the vaporized meth-
anol and reaction products. A thermowell in the center of the reac-
tor was used for measurement of the temperature. Three trefoil-
shaped baffles were welded on the outside wall of the thermowell to
reduce the gas bubble size.

Gaseous products exited the reactor through a side line at the

top of the reactor, and a cyclone was used to separate the fine cat-
alyst particles. Two traps in an ice case were used to collect
water, unreacted methanol, and other condensable products. The un-
condensed gas passed through a saturator at room temperature and then
through a wet-test meter to vent. Gas samples were taken right after
the cyclone and analyzed with a Carle GC Model 157 controlled by a
Hewlett-Packard Data Automation System [40]. Liquid products collec-
ted from the traps were weighed and then analyzed in a Varian Ge.
Catalyst particles were continuously fed and withdrawn during opera-
tion. A plate valve on the catalyst exist line was used to keep the
catalyst exit flow rate the same as the feed rate and to control the
catalyst residence time. Pressure drop measured with a manometer was
used to monitor the fluid bed height.

The coked catalyst was regenerated at 823°K with air while con-
tinuously feeding and withdrawing catalyst.

A typical catalyst particle distribution and density are shown

in Table 2.


';; Cyclone

Re<Jctor . ~




:>--_10 the Trops

MeIer Pump Ca talysl Collector

Used Catalyst
for Regeration

Figure 1. Schematic of fluidized bed reactor system.

Several types of experiments were conducted. The effects of

nitrogen, water, and C02 partial pressures on yield were studied.
Experiments were conducted by starting with a totally regenerated
catalyst and using it for two or more experiments without regenera-
tion. Experiments were conducted where catalyst regeneration resi-
dence times were equal to catalyst residence time when feeding meth-
anol. For a selected number of experiments coke content was measured
before and after the experiment.
336 L. LlU ET AL.

Table 2. Catalyst Size Distribution and Densitya

Size Undersize (by wt)

(l1IIl) Composition % Cumulative %

106 7.01 7.01

149 18.58 25.59
250 32.34 57.93
420 42.07 100.00

aDensity 0.67 g/cc; d p 173 microns


Effect of Catalyst Residence Times

A series of experiments were conducted at 703, 723, and 753°K

with a methanol weight hourly space velocity of 0.4 glhr methanol
per gram of catalyst. Figure 2 illustrates the advantage of using
a fluid bed unit with continuous catalyst flow. The time required
to achieve steady state is approximately the value predicted for a
well mixed reactor; (i.e. three times Sc' which is proportional to
the catalyst residence time). The catalyst residence times at 703°K
were 65, 55, 30, and 37 minutes.

Table 3 illustrates the effect of catalyst residence times and

catalyst regeneration on the steady state product distribution. Runs
324 and 329 are essentially replicas of each other as catalyst age
based on residence times is approximately equal (i.e. 55 min and 30 +
37 = 67 min). The major difference in the four experiments is Run
328, which illustrates the effect of the shorter catalyst residence
time on the product distribution. Operation at short catalyst res-
idence times resulted in a higher yield of propane and CS' thereby
reducing the yield to ethylene and propylene.

Figure 3 illustrates the effect of temperature and space veloc-

ity on the steady state hydrocarbon distribution. The primary effect
of increasing temperatures, as expected, is increased production of
methane. At higher space velocities, conversion decreased and the
ethylene yield increased. The weight percent of CO and C02 in the
total product increased from 1.2 to 5.4 with increasing temperature
(see later discussion of Figure 16) and the estimated carbon loss to
coke was 9, 18, and 19%, respectively, for MWHSV = 0.4.

A series of experiments were conducted to study the effect of

dilution of the methanol feed by nitrogen and by water. Ethylene
yields have been reported to increase with water dilution. At the
same time, the effect of catalyst age or time-on-stream on the yields
RltI'324 TEMP'430OC
CAT. RES. TIME- 55 min.
CAT. AGE' Regenerated with cotalyst re.idence
time of approximately one hour.

'11 0
~30 0


.8... "
"---0 Q

810 V

'0 CH4
FE ~ 8 Cfu ;

30 60 90 120 150 180 210

Time on Stream - Min.

Figure 2. Hydrocarbon distribution (steady state values are obtained

after 2.5 to 3.0 catalyst residence times).

was apparent. In order to study these two effects simultaneously,

a series of experiments were conducted by using the following proce-
dure. Total weight hourly space velocity, and hence Wc/VT' was main-
tained at essentially a constant value. Several steady state exper-
iments were conducted without regeneration between each experiment.
Difficulties were encountered in feeding the catalyst at the same
rate from experiment to experiment. Therefore, steady state catalyst
residence times were determined after an experiment by calculating
the average flow rate from the catalyst inventory in the feed tank
before and after the experiment. Once the catalyst flow rate for an
experiment was started, it could be maintained at a fairly constant
rate. The catalyst age was defined for these experiments as the sum
of the steady state catalyst residence times, 6c ' For example, the
catalyst was totally regenerated before Run 307, 6c = 20 min; no cat-
alyst regeneration after Run 308, 6c 30 min; therefore

50 min, and

[0N]307 = 20 min, etc.

The effect of the catalyst residence time on the ethylene con-

tent of the hydrocarbons distribution, as shown by Figure 4, was the
same whether it was achieved by making multiple passes through the
reactor without regeneration or by regeneration and conducting the
experiment at a longer residence time, 0 c • The experiments 323, 324,
328, and 329 were regenerated by using catalyst residence times ap-
proximately equal to the residence times used during the feeding of
338 L. LlU ET AL.

Table 3. Effect of Catalyst Residence Times and Regenerationa on

Product Distributions

Run No. 323 324 328 329

Temp. , °c 430 430 430 430
WHSV 0.4 0.4 0.4 0.4
% Conversion 100 100 100 98
Catalyst Residence
time, min 65 55 30 37
C2+C1 57 56 42 57.4
wt% 41.5 42 38.5 41.3
CO+C02, wt % 1.18 1.4 2.9 3.2
CH4 5.2 5.1 5.7 6.42
C2R4 30.2 29.7 26 29.8
C2H6 2.4 2.3 2.3 2.17
C3 H6 32.5 31.0 26.6 32.1
C3 H8 17 .6 16.5 20.35 14.5
C4 H10 4.05 4.02 5.09 3.31
C4 H8 6.3 9.88 9.45 8.8
C4 H6 0.05 0.05 0.05 0.06
C5 2.47 2.25 4.25 2.55

aCondition of the Catalysis: The catalyst for Runs 323, 324, and 328
was partially regenerated. Catalyst residence time during regen-
eration was approximately 60 min. Coke content was approximately
3.9 g of coke per 100 g of coke-free catalyst after regeneration
and before a run. Run 329: The catalyst was not regenerated after
Run 328.
bSelectivity C2H4 and C3H6 =100 (2 moles C2H4 + 3 moles C3H6)/(moles
methanol reacted) and reflects methanol conversion to coke, CO and
cMaximum theoretical value is 43.5% (Values reported are on a coke-
free basis).

reactants. Surprisingly, the effect of dilution by water or nitrogen

on the ethylene content of the hydrocarbons was not significant, and
the effect of catalyst residence time was negligible between 50 and
120 minutes; at eN = 155 min, DME-methanol conversion decreased to
70 percent.

The effect of dilution and catalyst residence time on the pro-

pylene hydrocarbon content, as illustrated by Figure 5, is more sig-
nificant than for ethylene. An increase in the mole fraction of

FEED -100% CH.DH FEED -100% C~OH

40 MeOH a DME CONY. 100%

5u 10


0L--4~30--------4~~--------~~~O~ OL--430~------4-5LO--------~~~~

Figure 3. Effect of temperature and space velocity on hydrocarbon


methanol in the feed from 12% to 48% results in an increase in the

propylene concentration. The propane content of the hydrocarbon
product, as shown in Figure 6, decreases with longer accumulative
residence times and with decreasing feed dilution. Since propylene
increases, propane decreases, and ethylene is essentially constant
'with an increase in the mole fraction of methanol in the feed, the
sum of the propylene, propane, and ethylene contents is essentially
independent of dilution and catalyst residence time as shown in Figure
7. The effect of dilution and catalyst age on the remaining hydro-
carbon products, butane, butylenes, ethane, and methane are shown in
Figures 8 through 11. Butadiene and C5 concentrations were less than
1% and 2%, respectively. The butane and ethane content increases
with increasing dilution. Butylene concentrations are higher for pure
methanol and for water dilution than for nitrogen dilution. Methane
concentration is less for pure methanol and water-methanol feeds.
Water or nitrogen in the feed appears to have the largest effect on
butylene and methane content (Figures 9 and 11) and on the CO and
C02 yield. Figure 12 illustrates the reduction in carbon loss as CO
and C02 for the three types of feeds. Distributions for Runs 323,
324, 328, and 329 do not always follow the trends discussed above,
but the gas residence times in the reactor for these experiments are
twice those of the experiments shown in Figures 4 through 12.

Figures 3 through 11 and Table 4 show that at high conversion

of methanol and dimethylether, the effect of dilution on the hydro-
carbon distribution is significant. The hydrocarbon yield is slightly
higher when water is the diluent instead of nitrogen (Table 4a). The
use of pure methanol results in a higher conversion of methanol to
coke. Failure to close the carbon balance by 10% is approximately
equivalent to four grams of coke per hundred grams of methanol re-
340 L. LlU ET AL.

.0 312

8 30
!' Feed Composition, mol. %
c 0
0._ ~OH :& EXP. NO.
II 0 0.' III 71 507,108,_.310 N-',I,2,5
>- 0 0.' 4. 51 SI',III,IIS N-',I,'
~ C, 0.1 100 !l9 N-I
W 10 'J 0.8 eo eo 517.'18 N = 1,2

~ 0 0.4 100 52S. S24,S28,H8 N-',',I,2

10 30 50 70 90 110 130 150 170
Accumulated Catalyst Residence Time (min.) SN" .-Z:1 8..

Figure 4. The effect of catalyst activity and feed composition on

ethylene yields, T = 703°K (Methanol-DME conversion was
greater than 95% in all experiments except 313 where it
was 70%).

acted. The major effect of dilution is to decrease the rate of

coking (Table 4b), and the major effect of water is to decrease the
amount of DME, CO, and C02 produced, and thus increase the hydro car-
bqn yield per gram of methanol reacted (Table 4c). At high methanol

Feed Composition, mole %

12. lOe,lOa N.',2
ae 71 507,101,301,510 N-',',2,S
0.8 100 ,.8 N-I
0.8 50 eo ~7,11' N ~1,2
0.4 100 I2I.U4,11',H8 N-',I,',2

~ 30 50 70 90 00 00 ~ ~
Accumulated Catalyst Residence Time (min.) SN" .-Z:1 8..

Figure 5. The effect of catalyst activity and feed composition on

propylene yields, T = 703°K (Methanol-DME conversion was
greater than 95% in all experiments except 313 where it
was 70%).

'E o
Feed Composition, mole %
~OH :. HaD =.:. •. t,2

5 40 DO.•
o 0..
!01,!08.1C8,1fO N-',',2.!
511,111,111 N-',I,I

c I:> 0 .•
'\} 0..
o 0.4
N • ',2
!2!,524,la• • • Ncl,I,I,2
.5 20
e 10 - ......."'-- 25%

10 30 50 70 90 110 130 150 170

Accumulated Catalyst Residence Time (min.) • 8..

eN" .7:

Figure 6. The effect of catalyst activity and feed composition on

propane yields, T = 703°K (Methanol~DME conversion was
greater than 95% in all experiments except 313 where it
was 70%).

conversions, where all of the dimethylether is reacted, the effect

of nitrogen versus water dilution is less significant. However, at
low conversions and high space velocities the use of water as the
diluent significantly increases the yields.


.t~ >Ie I'
+1: 323 V
!i 80 <>.306
0 31•

!e Feed Composition, mol. %
o D.. ~OH : . HaD !:;:. •'1,2
£.E 60 o 0.. ae TI "',101,_.510 N-',',.,I
1&1 o 0.. 41
al 111,.11.51....1, •••
.. ._,
I:> 0 .•
~ '\} 0.1 DO eo 517,118 It -',2
o 0.4 100 125.124.11',121 N-',I,I,2
10 30 50 70 90 110 130 150 170

Accumulated Catalyst Residence TIme (min.) eN".~ 800 •

Figure 7. The effect of catalyst activity and feed composition on
ethylene, propylene, and propane yields, T = 703°K (Meth~
anol~DME conversion was greater than 95% in all experi~
ments except 313 where it was 70%).
342 L. LlU ET AL.

.ti 10
FBBd Composition, mole %
: . HrfJ
::;:e N'I,2
507,108.101,310 H-',',2,5
1ii 8 0
0 0.. 48 SI 511,112.115 "- •• 2.1
0 o.a

{> 100 !f9 N~I
c 'V 0.8 SO 50 117,118 N =1.2
11 <> 0.4 100 311.114,128,128' N-',I,',2

8 6

.E 4

GI 2
10 30 50 70 90 110 130
__ __
Accumulated CatalYlt Relidence Time (min.! eN"!.. 8..

Figure 8. The effect of catalyst activity and feed composition on

butane yields, T = 703°K.

The effect of carbon dioxide as the diluent was studied by op-

erating the reactor without continuously feeding and withdrawing the
catalyst, i.e. batch catalyst. Tables 5a and 5b show product distri-
butions for C02 dilution, and these can be compared with the distri-
butions given in Tables 6a and 6b for water dilution. The major dif-
ferences are again in the rate of coking, as indicated by lack of
closure of the carbon balances and on the production of CO and C02.

10 ...
48% MIN


FBBd COmposition, mole % 0315

D a.. ~OH :. 5OB,"
0 0.' g 78 !07,IOI,IOI,SlO
0 0.. 4. 51 SlI.5I1.111
{> 0.' 100 II'
'V 0.8 !SO 50
<> 0.4 100 511, U4.SI'.HI

30 50 70 90 110 130 150 170

Accumulated CatalYlt R..idence Time (min.! eN" .~18..

Figure 9. The effect of catalyst activity and feed composition on

butylene yields, T = 703°K.

c Feed Composition, mol. %
~OH ~~,:. ""',e
0.8 :. H,.o
o 0 .• 28 75 !07,508,3OI,atO Na',1,2,1
8 o 0.. 41 81 III,SI2,all N -',.,3
C D. 0.9 100 319 Na,
'J 0.8 50 80 317,348 N =1,2
c o 0.4 100 321, 324, a2~,528 He ',1,',2
~0 6
.E 4 /12%, 25and48% MINe
~7 305 I
306 309 312
C j38 0311 ~13
.c o~ ~K>
Ui 2 6319~ ..... ,
•:.e 100 % M and 50-50 MIW
10 30 50 70 90 110 130 150 170

Accumulated Catalyst Residence Time (min.) eN" .~, 8.. "

Figure 10. The effect of catalyst activity and feed composition on
ethane yields, T = 703°K.

The maximum ethylene yields are obtained in a shorter time-on-stream

for C02 than for water.

The hydrocarbon product distribution at 703, 723, and 753°K and

WHSV of 0.4, 0.94, and 2.0 for 100% methanol feed was correlated as
a function of propane content as shown in Figure 13. As expected,
the maximum. yield of ethylene and propylene occurs at a low propane
content. Figure 14 shows the decrease in propane content as the

c 0'" 48-52 M-N
.8 _ _-=-----~00313
~ 6
.E 4 2 317 SO-50 M-W
• 6319 n 318_ Feed Composition, mole %
c WHOY ....OH ". . . . "P. NO.
~ 0 0.8 12 81 508,301
; 2 0 0.8 ae 71 !O7,50.,_,310
~ 0 0.. 48 51 ''',31e,31'

g: ':
0.4 100
50 ~I~,al.

10 30 70 90 110 130 150 170
Accumulated Catalyst Residence Time (min.) eN" .~t 8..

Figure 11. The effect of catalyst activity and feed composition on

methane yields, T = 703°K.
344 L. LlU ET AL.



110 150 170

Accumulated Catalyst Residence TIme (min.) eN" .'1!:1 9..

Figure 12. The effect of catalyst activity and feed composition on

CO + C02 yields, T = 703°K.

space velocity increases for the temperature of 703°K. The decrease

in olefin content as paraffins increase is shown in Figure 15. The
effect of temperature and space velocity on the product distribution
is illustrated in Figure 16.

The rate of coke production based on "burnout" studies correla-

ted well with the closure of the elemental carbon balance. As shown
in Table 7 and Figure 17, low closure of the carbon balance corre-
sponds to high rates of coking. Therefore, lack of carbon balance
closure is an excellent measure of carbon loss by coking. Figure 17
indicates a stoichiometric relationship between coke production and
paraffin yield. The coke production, as indicated in Figure 18 by
a lack of closure of the carbon balance, was dependent on space
velocity and temperature.


The interpretation of data from a fluidized bed reactor is dif-

ficult because of insufficient knowledge of the flow patterns. Since
two and a half to three catalyst residence times, defined as the
ratio of mass of catalyst in the reactor to mass feed rate of cata-
lyst, were required to reach steady state, the catalyst bed flow pat-
terns appear to be characterized as "well mixed". Gas residence
times were less than 5 seconds, and the degree of back mixing of the
gas is unknown. Product distributions are similar to data obtained
by using tubular reactors that are reported in the literature [9, 13,
15, 17] and in Table 1. Thus, the gas appears to be nearly in plug
flow. Yields are substantially higher than values reported by Chang
Table 4a. The Effect of Nitrogen and Water Dilution on Hydrocarbon Yields z
Product Yie1da m
W, ]J.g/ SELEC Coke-Free Basis J:
Ci & C3 CO & CO2
Wt % 0
Wt % Wt % r
323 0.4 101.3 100 57 42.42 1.22 0
100% CH30H :E
324 100% CH30H 0.4 101.3 100 56 41.5 1.23 s:::
311 50% CH30H 0.45 42.19 98 61.2 42.8 2.4 0
50% N2 m
318 50% CH30H 0.52 38.9 95 63.0 43.7 0.33 cr
50% H2O »
319 100% CH30H 0.94 45.76 94 49 42.27 0.43 1.07 ::0
333 50% CH30H 0.93 21.75 55 28.29 18.0 2.29 42.6 r