Chemical processes in viscous materials

MISSING IMAGE

Material Information

Title:
Chemical processes in viscous materials
Physical Description:
xi, 137 leaves : ill. ; 28 cm.
Language:
English
Creator:
Barnes, Mark Jay, 1962-
Publication Date:

Subjects

Subjects / Keywords:
Membranes (Technology)   ( lcsh )
Polymers   ( lcsh )
Transition metal catalysts   ( lcsh )
Catalysis   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1989.
Bibliography:
Includes bibliographical references (leaves 130-136)
Statement of Responsibility:
by Mark Jay Barnes.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001512996
oclc - 21924307
notis - AHC5987
sobekcm - AA00004774_00001
System ID:
AA00004774:00001

Full Text











CHEMICAL PROCESSES IN VISCOUS MATERIALS


BY

MARK JAY BARNES















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1989























TO MY FAMILY WHOSE SUPPORT NEVER FALTERED













ACKNOWLEDGEMENTS


The most cherished and remembered things which a person can obtain

in his lifetime are those which require sacrifice and dedication to

achieve. The pursuit of knowledge is certainly no exception to this

definition. If exception be taken with this definition, it is due to

the implication that sacrifice and dedication are contributed only by

the individual. On the contrary, an education would be impossible to

achieve without the sacrifice and dedication of a large number of

people.

First and foremost I wish to express my gratitude to my family.

Their continued support and encouragement have given me the opportunity

to fulfill what was once a dream. My parents, Avery and Marilyn, have

instilled in me a sense of security which can never be taken away. From

them I have learned what happiness is, from where it comes, and how it

can be obtained. To my brothers and sister-in-law, Douglas, Curtis, and

Robin, I owe my love of science. It was they who pointed me in the

right direction and taught me to never be apprehensive of the

unexpected. To my grandparents I am indebted for all their generosity

and support.

I would also like to thank my teachers who have instilled in me a

love for chemistry, Glenn Vogel, Heinz and Judy Koch, Harry Sisler, and

most importantly Russell "Doc" Drago. It is truly an inspiration to








know a person who loves and enjoys his work as much as he does. I would

also like to thank Ruth Drago for making Florida my home away from home.

I would also like to thank all of my friends. There are too many

to list, but there are those who have influenced me greatly in one way

or another. Included in this list are the past and present residents of

the "Thunderdome": Todd "Spud" Gillespie, Stephen "Julio" Brooks, Mark

"Chin" Hail, Gerald "Citrus Connection" Grunewald, and Donald "Bueller"

Ferris. I would also like to extend my gratitude to past and present

members of the Drago Group, in particular, Maribel Lisk, Robert Taylor,

Larry Chamusco, Edward Getty, Cindy Getty, Shannon Davis, Richard Riley,

and especially Ngai Wong, who has been a good friend and source of

information for many years. Lastly, I would like to express my

gratitude to my girl friend, Kaneez Rizvi. I owe a considerable portion

of my sanity to her for the love and understanding she has given me.














TABLE OF CONTENTS

page

ACKNOWLEDGEMENTS................................................ iii

LIST OF TABLES................................................... vi

LIST OF FIGURES.................................................. viii

ABSTRACT ......................................................... x

CHAPTERS

I GENERAL INTRODUCTION....................................... 1

II COBALT(II)-FACILITATED TRANSPORT OF DIOXYGEN IN A
POLYSTYRENE MEMBRANE....................................... 4

Introduction............................................... 4
Experimental............................................... 16
Results and Discussion..................................... 21

III DEMONSTRATION OF THE FEASIBILITY AND VERSATILITY OF
MEMBRANE REACTORS.......................................... 43

Introduction............................................... 43
Experimental .............................................. 58
Results and Discussion..................................... 62

IV HETEROGENEOUS HYDROFORMYLATION OF PROPYLENE USING SUPPORTED
RHODIUM CATALYSTS IN A CONTINUOUS GAS FLOW REACTOR......... 80

Introduction............................................... 80
Experimental............................................... 87
Results and Discussion..................................... 91

V SUMMARY.................................................... 115

APPENDIX

COMPUTER PROGRAM......... ................................. 117

REFERENCES ... .................................................. 130

BIOGRAPHICAL SKETCH ........................................... 137

v













LIST OF TABLES

Table page

2-1. Permeation data for PS/[P]-CoSDPT (16.4%)
membrane... .................... ................ 22

2-2. Separation data for a series of polystyrene
blank membranes................................ 27

2-3. Separation data for a series of nickel blank
membranes..................................... 28

2-4. Difference in oxygen partial pressure for a
series of PS/[P]-CoSDPT membranes.............. 32

2-5. Difference in oxygen partial pressure for a
series of PS/[P]-CoBr2SDPT membranes........... 33

2-6. Difference in oxygen partial pressure for a
series of PS/[P]-Co3FSDPT membranes............ 34

2-7. Difference in oxygen partial pressure for a
series of PS/[SG]-CoSDPT membranes............. 35

2-8. Comparison of the difference in oxygen partial
pressures obtained with increased cobalt
loading in PS/[P]-CoSDPT ....................... 38

3-1. Hydroformylation of propylene with
RhH(CO)(PPh3)3................................. 66

3-2. Oxidation of 1-butene......................... 67

3-3. Butyl sulfide membrane conversion data for
various ruthenium catalyzed oxidations by 02 at
1 atm. and 900C................................ 71

3-4. Butyl sulfide membrane turnover data........... 72

3-5. Butyl sulfide solution reaction conversion data
for various ruthenium catalyzed oxidations by
02 at 900C .................................... 74

3-6. Hydrogen peroxide formation .................... 78








4-1. Temperature effect upon propylene
hydroformylation activity using RhH\SG......... 94

4-2. Selectivity as a function of time in the
hydroformylation of propylene with RhH\SG...... 96

4-3. Effect of gas flow on the hydroformylation of
propylene with RhH\SG.......................... 100

4-4. Effect of gas pressure on the hydroformylation
of propylene with RhH\SG....................... 103

4-5. Effect of gas ratios on the hydroformylation of
propylene with RhH\SG.......................... 105

4-6. Catalyst comparison at 800C.................... 109

4-7. Catalyst comparison at 1000C................... 112













LIST OF FIGURES

Figure page

2-1. Molecular orbital diagram.
a) Dioxygen; b) Dioxygen adduct of cobalt(II).. 7

2-2. Polystyrene supported N,N'-bis(salicylidene-
imino)di-n-propylamine cobalt(II), [P]-CoSDPT.. 10

2-3. Schematic representation of the dual mode
sorption concept............................... 13

2-4. [P]-CoSDPT reaction scheme..................... 19

2-5. Schematic diagram of the gas permeation
experimental set up............................ 20

2-6. Plot of the natural log of oxygen partial
pressure differential as a function of time
for a PS/[P]-CoSDPT, (16.4%), membrane......... 24

2-7. Plot of the natural log of nitrogen partial
pressure differential as a function of time
for a PS/[P]-CoSDPT, (16.4%), membrane......... 25

2-8. Oxygen partial pressure as a function of time.
a) Experimental curve. b) Calculated
non-facilitated curve. c) Calculated
reference curve (a = 1.00)..................... 30

2-9. Contributions to permeation mechanism.......... 40

2-10. Schematic representation of a metal facilitated
transport mechanism due to site to site
interaction.................................... 41

3-1. Read and Dudley mechanism for the oxidation of
terminal alkenes with RhH(CO)(PPh3)3........... 49

3-2. Proposed mechanism for the oxidation of
terminal alkenes using a
rhodium(III)/copper(II) co-catalyst............ 51


viii








3-3. Alternative mechanism for the oxidation of
terminal alkenes using a
rhodium(III)/copper(II) co-catalyst............ 52

3-4. Sulfide oxidation catalysts:
a) [Ru(O)2(dmp) ](PF6) and
b) [Ru30(prop)6 H20)3] ........................ 55

3-5. Anthraquinone autoxidation process............. 56

3-6. Reactor membrane apparatus..................... 61

3-7. Hydrogen peroxide membrane reactor............. 76

4-1. Associative and dissociative hydroformylation
reaction mechanism............................. 84

4-2. Gas flow reactor schematic..................... 90

4-3. n-Butanal activity as a function of temperature
in the hydroformylation of propylene with
RhH\SG......................................... 95

4-4. Comparison of hydroformylation activity in the
production of n-butanal and isobutanal with
RhH\SG......................................... 97

4-5. Effect of gas flow on the n-butanal activity in
the hydroformylation of propylene with RhH\SG.. 99

4-6. Effect of gas pressure on the n-butanal
activity in the hydroformylation of propylene
with RhH\SG.. ................................. 102

4-7. Effect of gas ratios on the n-butanal activity
in the hydroformylation of propylene with
RhH\SG........................................ 104

4-8. n-Butanal activity for RhH\SG and Rh(tfa)\SG... 107

4-9. Butanal activity as a function of time in the
hydroformylation of propylene with RhH\SG\PC at
100C .......................................... 113














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

CHEMICAL PROCESSES IN VISCOUS MATERIALS

By

Mark Jay Barnes

August 1989

Chairman: Russell S. Drago
Major Department: Chemistry


Polymer membranes, in which transition metal complexes have been

dispersed, have been used to facilitate both oxygen enrichment of air

and traditional homogeneous catalytic reactions. The former is a

demonstration of facilitated transport of oxygen through a glassy

polymer. The latter includes the demonstration of a wide variety of

reactions within the polymer environment.

Selective gas permeation through a polymer membrane has become an

attractive alternative for conventional gas separation processes. The

incorporation of additives that modify permeation properties has been

the focus of this research effort in an attempt to enhance the

separation ability of the polymer membranes. This effort involves the

use of metal complexes to facilitate the transport of gases in a polymer

membrane. Specifically, the main objectives investigated were the

incorporation of supported cobalt (II) schiff base complexes into

polystyrene membranes and the demonstration of metal promoted








enhancement of oxygen permeation in the separation of air. Back

calculation using a nickel blank to provide an average separation factor

allows for the determination of metal promoted enrichment using the

nitrogen permeability coefficient as an internal standard. These

results are presented and a transport mechanism discussed.

The use of a polymer environment as the medium for chemical

reactions has resulted in a novel approach to heterogenizing homogeneous

catalytic reactions. The feasibility of this process has been

demonstrated by the hydroformylation of propylene and the oxidation of

terminal alkenes. A specific application of the process involves the

use of ruthenium catalysts dispersed in composite polymer membranes for

the oxidative decomposition of butyl sulfide, a mustard agent simulant.

The formation of hydrogen peroxide using the anthraquinone autoxidation

process is another application which has been achieved. Results from

the above mentioned reactions are presented and discussed.

The development of solid supported rhodium catalysts for

heterogeneous hydroformylation reactions has resulted in the production

of straight chain aldehydes with high selectivity. Use of these

catalysts in a pressure-gas flow reactor is demonstrated and discussed.













CHAPTER I
GENERAL INTRODUCTION


Historically, applications of polymer membrane technology have

been dominated by separation processes. The two biggest contributors to

this field are reverse osmosis desalination processes and industrial gas

separations. Many smaller scale applications which utilize membrane

technology are currently being investigated.1'2 Drug production and

delivery, fermentation, basic chemicals synthesis, hydrogenations and

dehydrogenations represent only a portion of these applications. The

types of membranes employed are as widely varied as their applications.

Ceramic, metallic, glass, inorganic, and polymeric membranes have been

used as process media.3-6 Also of interest is the engineering design in

which they are used. Integrated or "hybrid" processes are being

developed which combine a membrane separation unit with an existing

separation process such as cryogenic separation.7,8 Layering or

combining the different membrane types may provide systems which allow

chemical reaction and product-reactant separation within the same unit.

The work presented in this dissertation is associated with

combining membrane technology and transition metal reaction chemistry.

The use of polymers and catalysts in conjunction has previously been

widely investigated in an attempt to heterogenize known homogeneous

catalysts.912 For the most part, the polymer is used as a support to

which the catalyst is covalently attached. These insoluble








2

functionalized catalysts are then used under "normal" solution chemistry

conditions. Alternatively, membrane reactors are systems in which the

polymer is used as the medium for reaction. The polymer functions, in

essence, as the solvent for the reaction.

The concept of membrane reactors is relatively new. Wolynic and

co-workers first published work in this area which dealt with a

catalytic liquid membrane reactor used for the production of

acetaldehyde from ethylene and oxygen.13,14 Since that time most

membrane reactor work has been concentrated in two areas. Membrane

bioreactors have been shown to be effective in enzyme catalysis2,15-17

and porous ceramic and glass membranes have been used as inorganic

membrane reactors.3-6 In the latter case, hydrogenation and

dehydrogenation reactions are the main processes being attempted.

Catalysis with nation membranes has also been reported.18,19 Nafion

possesses a chemically resistant, perfluorinated, polymeric backbone

with highly acidic sulfonic acid groups. Its general structure is shown

below. It has been used as a super acid catalyst, catalyst support,

electrocatalyst, and a gas diffusion membrane.



[(CF2-CF2)n-CF-CF2]x

(OCF2CF)mOCF2CF2SO3H

CF3



Even though a considerable amount of work has been conducted in

the area of membrane reactors, little of it has actually been achieved








3

with the use of organic polymers. The work presented here is an attempt

at filling this void. Specifically, this study involves three areas of

membrane chemistry. The first process investigates the ability of a

transition metal complex to affect the permeation properties of a

polymeric membrane. The second is concerned with how transition metal

complex reactivities are influenced by dispersion in an organic polymer

"solvent" environment. The last area examined deals with dispersing

transition metal complexes in a non-volatile viscous material which is

coated on an inorganic oxide support for the purpose of creating a

heterogeneous gas phase reaction process.













CHAPTER II
COBALT(II)-FACILITATED TRANSPORT OF DIOXYGEN
IN A POLYSTYRENE MEMBRANE


Introduction


The separation of air into its primary components, oxygen and

nitrogen, is of major industrial significance. Today, two main

commercial processes are used to achieve separation, cryogenic

fractional distillation of air20-22 and pressure swing adsorption.23-25

Both processes have been employed extensively in the last century for

the production of high purity oxygen and nitrogen.

Many applications, however, do not require high purity oxygen but

can instead use oxygen enriched air. This market includes waste water

treatment facilities, the pulp and paper industry, fermentation

processes and numerous medical applications to name a few. Selective

gas permeation2528 through a polymer membrane provides an alternative

separation method which is becoming increasingly more important in the

production of oxygen enriched air.29'30

The past ten years have seen the advent of membrane based

separation units. The first commercial venture into the field was

initiated by the Monsanto Corporation. It utilized a polysulfone hollow

fiber membrane system and was called "The Prism Process."29 Since that

time many types of membranes have been increasingly employed for

separation purposes, and recently, are being used in conjunction with










the traditional air separation processes.31,32 These hybrid systems

have been shown to be an economically attractive alternative in the

production of pure oxygen and nitrogen. The fact that membrane

separation units are economically competitive with existing processes,

coupled with the unlimited number of applications, has resulted in an

explosion in the amount of research being carried out in this field.

Interest is mainly twofold. Most attention is focused on finding new

applications for membrane separation units and developing new, or

modifying existing, polymers for these applications.The use of membranes

for gas separations is based upon the phenomenon that different

permeates will have varying rates of permeation through membranes. One

of the first references to a membrane separation procedure is credited

to Thomas Graham,33 who, in 1864, used a dialyzer to separate a solution

into its components. He became known as the father of gas separations

when he published his famous paper "On the Molecular Mobility of

Gases"34 in 1863, which resulted in the formulation of Graham's Law, and

later published a paper dealing with the separation of gases.35 Since

that time many others have made considerable contributions in the areas

of membrane chemistry and gas separations, but because of the size and

scope of this work, it will not be mentioned in this chapter.

Alternately, a more detailed insight into the properties which affect

membrane permeation will be presented.

Ideally, a permselective polymer membrane should have good

mechanical strength, high selectivity, and high flux (rate of gas

permeation). However, the later two are inversely related. A polymer

membrane which has a high selectivity will have a relatively slow flux










and vice versa. Attempts to either increase the rate of permeation in

highly selective membranes or enhance the selectivity of highly

permeable membranes have been the focal point of most research efforts.

The work presented in this chapter represents an undertaking in the

latter area of modification.

The incorporation of additives to modify permeation properties has

been directed towards improving membranes that are highly permeable and

poorly selective. A recent approach to this concept involved enhancing

the transport of dioxygen by encapsulating a metal complex-containing

solution in a porous polymer membrane resulting in excellent

selectivity.3639 This liquid-membrane, oxygen-enrichment process

utilizes cobalt(II) chelate complexes. The role of the cobalt complex

is to act as a carrier by reversibly binding dioxygen and thus increase

the oxygen permeability of the membrane. In systems such as these, the

most important aspects to be considered are the processes which occur at

the carrier site. Essentially, these processes are attributed to the

reversible binding ability of the cobalt(II) complex and can be

explained by extrapolating our knowledge of similar processes which

occur with cobalt(II) complexes in solution.

The ability of the metal center to bind dioxygen and the factors

which affect the interaction between the two can be complex. As shown

in figure 2-la, molecular oxygen in its ground state is a triplet, 3 -

molecule with two unpaired electrons residing in the 7 molecular

orbitals. Since it is kinetically unfavorable for one of the electrons

to change spin, dioxygen is typically unreactive towards diamagnetic

compounds. Its ability to bind with transition metal complexes results











iL-i->'


3ou *

\\ '1 _1 L


\\L /L I
3ag


2\u
IL 2ou*
- .
S .- 2g

02 0


b) 3dx2-y2


l-,


- r-- t
- -AL --


AL -4
--- -___IL


ALE--1


Co


Figure 2-1.


Molecular orbital diagram.
a) Dioxygen; b) Dioxygen adduct of cobalt(II).


3dz2


3dxy


3dxz, dyz


l 7g *


0A

Co
/I\


ILIL
11


ITu
3ag










primarily from spin orbital coupling, which lessens the severity of a

spin change. The spin pairing model40 developed by Drago and Corden

describes this metal-dioxygen bonding scheme and is shown in figure 2-lb

using a simplified molecular orbital diagram. It involves the pairing

of the unpaired dz2 electron of cobalt(II) with an unpaired 7r electron

of dioxygen which results in the formation of an end on bonded

cobalt(II)-dioxygen adduct.

The stability of the cobalt-oxygen sigma bond is directly related

to the ligand field strength of the complex. By increasing the ligand

field strength around the cobalt metal center, the energy of the dz2

orbital used in bond formation is increased. Therefore during dioxygen

adduct formation, more energy is given off resulting in a more stable

cobalt-oxygen sigma bond. This is a critical property since if the bond

formed is too stable, the ability to dissociate the dioxygen adduct is

hindered. If the bond formed is too weak, the complex will have poor

oxygen affinity. Thus a compromise is required to have a suitable

carrier.

In 1938, Tsumaki reported the first synthetic complex capable of

reversibly binding dioxygen, N,N'-bis(salicylidene)ethylenediamine

cobalt(II).41 This complex, better known as cobalt(II) salen, is shown

below. Since that time, many cobalt(II) schiff base complexes have been










reported which are capable of reversibly binding dioxygen.

Unfortunately, deactivation of the metal complexes has hindered their

development. Deactivation, for the most part, is brought about by

either oxidation of the ligand system or irreversible oxidation of the

metal center. An example of the latter would be the formation of a

A-peroxodimer complex. Attempts to achieve a stable oxygen carrier have

been directed using two different approaches. Sterically constrained

metal carrier complexes and supported metal carrier complexes have

received the greatest attention. Often, schiff base complexes which

indicate promise as carriers are the focus of such modification. This

is the case in the work presented in this chapter.

One of the best examples cited which has the ability to bind

dioxygen reversibly is N,N'-bis(salicylideneimino)di-n-propylamine

cobalt(II), CoSDPT (figure 2-2 with P replaced by H), in 1:1 DMSO and

a-butyrolactone solvent encapsulated in a 130 p thick microporous nylon

6,6 membrane.42 At 25C, an air mixture containing 88% oxygen was

produced in a single pass through this membrane. A schematic diagram of

this system is illustrated below. Although the process appears quite


MEMBRANE
FEED STEAM PRODUCT STEAM
CARRIER
ATMOSPHERIC AIR LOWER 02 PRESSURE
:3 \OR
S/ HIGHER TEMPERATURE
02 02















































R = H, Br, F


Figure 2-2.


Polystyrene supported N,N'-bis(salicylideneimino)di-
n-propylamine cobalt(II), [P]-CoSDPT.








11

remarkable, the removal of the volatile solvent by gas flow has hindered

its commercial development. The work presented in this text illustrates

the ability to reversibly bind and facilitate the transport of dioxygen

through a polymer membrane using this complex supported on polystyrene.

A few reports of facilitated transport of dioxygen by carriers in

membranes have appeared43'44 using unsupported complexes that can be

viewed as being dissolved in the polymer.

The concept of a facilitated transport mechanism has been a

controversial subject in membrane separations. Therefore, a brief

mathematical description of permeation in a polymer membrane is

necessary. According to Fick's Law of diffusion, shown below, the flux,



J = DSAP/1



J, of a penetrant can be determined from its diffusion coefficient, D;

solubility, S; and pressure differential, P, across a membrane of

thickness 1. The permeability of the gas, P, is dependent upon both its

solubility and diffusivity in the respective membrane.



P = DS



Generally, sorption and transport in a glassy polymer such as

polystyrene is best explained by a dual mode model. Using this model,

two simultaneous processes in the membrane are predicted to occur. The

first is associated with dissolution in the dense region of the polymer

and follows Henry's Law where CD is the concentration, KD is the Henry's











CD = KDP



law constant which characterizes sorption of the penetrant, and P is the

pressure of the penetrant. The second process is associated with

distribution in the gaps or "holes" of the polymer and is best explained

by a Langmuir isotherm where CH is the concentration, CH' the Langmuir


CH = CH'BP / (1 + BP)


sorption capacity, and B the affinity of the penetrant for the gaps in

the polymer. Using these simple equations, the total concentration of

the penetrant can be described as the sum of the two individual


C = CD + CH



concentrations. This can be seen schematically in the concentration

versus pressure graph shown in figure 2-3. The overall solubility

isotherm can similarly be described as the sum of the solubility

expressions of each model. The permeability of the penetrant


S = KD + CH'B / (1 + BP)


can now be expressed as the diffusivity and solubility associated with

each individual mode where DD and DH represent the diffusivities of the

Henry's mode and Langmuir mode respectively.


























k


p
P





..CH-....**_




P


Figure 2-3.


Schematic representation of the dual mode sorption
concept.


4-

C
H











P = KDDD + CH'BDH / (1 + BP)


The model presented is representative of a nonfacilitated process

and must therefore be modified to explain the contributions made by a

facilitated transport mode. This can be done by initially applying

Langmuir kinetics to describe the equilibrium obtained when oxygen binds

to solid metal complexes. The resulting expression can then be

rearranged to indicate the fraction of oxygenated sites, X, with


X / (1 X) = KP02


X = KP02 / (1 + KP02)


respect to the equilibrium constant, K, and dioxygen partial pressure.

The fraction of oxygenated sites should be directly related to the

solubility of dioxygen associated with these sites. Therefore, a new

solubility isotherm may be written to express the total solubility of

dioxygen in a membrane capable of facilitating dioxygen transport.

Similarly, a new expression describing dioxygen permeability according

to the three modes of transport can be formulated where DM represents

the diffusivity associated with the metal sites.


S = KD + [CH'B / (1 + BP)] + [KP / (1 + KP)]


P = KDDD + [CH'BDH / (1 + BP)] + [KPDM / (1 + KP)]










The results presented in this work represent a new approach to

enhancing the permeation properties of polymer membranes and

establishing a novel method and mechanism of transport. The goal of

this research was to demonstrate this novel mode of transport as well as

to develop a process that accurately accounted for the quantities of

dioxygen associated with both facilitated and nonfacilitated transport.

To achieve this, a metal complex that reversibly binds dioxygen is

covalently attached to polystyrene. Though the pure complexes in the

solid state do not bind dioxygen, they do bind dioxygen in the solid

state when covalently linked to a polymer.45 The dispersion of the

metal complex on highly crosslinked polystyrene into a polystyrene film

resulted in the selective permeation of dioxygen through the membrane.

This membrane system differs from the liquid membranes in that the metal

complexes are bound to the polymer. Since the cobalt complexes are not

mobile, the mechanism is unlike that proposed for isotropic polymers or

liquid membranes as demonstrated in the Bend Research system.42

Recently, this system has been extended to cobalt(II) porphyrin

and Schiff base complexes.46-49 In all reports43'44'46-51 of

facilitated transport in solid polymer membranes, the demonstration of

facilitated transport by metal complex binding of dioxygen is dependent

upon the critical selection of a blank. Incorporation of a material in

a polymer film can function to change the structure of the membrane and

the permeability of gases leading to enrichment by a process other than

facilitated transport. The results of this study are significant for

they provide strong support for a facilitated transfer mechanism using

metal complexes to bind simple gas molecules. A procedure for the










evaluation of the experimental permeation data is reported that

eliminates imperfections including solvent entrapment that surely exist

in the test and blank films.


Experimental


Complexes studied were prepared using reagents which were either

purchased or synthesized using the procedures listed below. An overall

reaction scheme is shown in figure 2-4 for the preparation of the

polystyrene supported complexes. Solvents used in this work were

purified by distillation, stored over 4A molecular sieves under a

nitrogen atmosphere, and degassed prior to use when necessary.

Characterization of the complexes was conducted using a Perkin Elmer

Plasma II Emission Spectrometer, and Bruker ER 200D-SRC Electron Spin

Resonance Spectrometer. Elemental analyses were conducted by the

University of Florida Department of Chemistry Microanalysis Service, and

Galbraith Laboratories.

Polystyrene Supported Dipropylenetriamine, [P1-DPT

Polystyrene beads (90% chloromethylated, 4% divinylbenzene

crosslinked) were donated by Sybron Corporation. Polystyrene bound

dipropylenetriamine was prepared according to the literature.45

Experimental analysis: %C = 67.41, %H = 6.64, %N = 7.60. Theoretical

analysis: %C = 72.83, %H = 10.19, %N = 16.98.

Polystyrene Supported Bis(3-(salicylideneamino)propyl)amine, [P1-SDPT

The polymer bound pentadentate ligand was prepared according to

the literature.45 The 3,5-dibromo and the 3-fluoro substituted










salicylaldehyde derivatives were prepared in a similar manner.

Experimental analysis: %C = 75.97, %H = 7.19, %N = 7.84. Theoretical

analysis: %C = 76.45, %H = 7.30, %N = 9.22.

FP1-CoSDPT, -Co Br2SDPT and -Co3FSDPT

Cobalt was incorporated into the [P]-SDPT type complexes by

slurrying excess cobalt(II) acetate with the appropriate resins in DMF

under argon at room temperature for two days. The polymer supported

cobalt complexes were filtered, washed with DMF and dried under vacuum

at 80C. Experimental analysis of [P]-CoSDPT: %Co = 0.15. Theoretical

analysis: %Co = 11.50. Overall analysis indicated greater than 95% of

the chloromethylated polystyrene had been converted to the [P]-SDPT and

that 1.15% of the SDPT ligand contained cobalt. The low loading of

cobalt was ideal for this study because site isolation inhibits

formation of the /-peroxo dimer. EPR studies showed an intense cobalt-

dioxygen signal which was used to monitor reversible binding of 02.

Silica Gel Supported CoSDPT, -CoBr2SDPT and -Co3FSDPT

Silica gel supported dipropylenetriamine was prepared as reported

in the literature.52 The silica supported Cobalt(II)SDPT complexes were

prepared from [SG]-DPT in an analogous manner to the polystyrene bound

analogues. Experimental analyses were not obtained.

[PI-NiSDPT, -NiBr2SDPT, and [SGI-NiSDPT

The analogous nickel complexes were prepared in the same manner as

reported for the cobalt complexes. Analyses were not obtained.

Membrane Preparation and Characterization

The membranes employed in this study were prepared using a casting

technique. Polystyrene was dissolved in an appropriate solvent, such as










toluene or methylene chloride. The viscous polymer solution was then

poured into a leveled circular form. Evaporation of solvent at room

temperature resulted in the formation of homogeneous polymer membranes.

Membranes containing the supported metal complexes were prepared

in a similar manner by grinding the complex into a fine powder (< 1/ in

diameter), adding it to the polymer solution, and then casting the

solution in a mold. After solvent evaporation had occurred, the

membrane's thickness was determined with a vernier caliper to the

nearest hundredth of a millimeter. The weight percent of the added

complex in the membrane was determined after each permeation experiment

by redissolving the polymer film, isolating and drying the supported

complex and then weighing it.

Permeation Apparatus

Permeation experiments were carried out with an apparatus designed

by Ken Balkus.50 The cell was designed to support the membrane and

prevent rupturing when a pressure gradient was applied to the membrane.

A schematic of the permeation setup is shown in figure 2-5.

Permeation Procedure

The following section describes a typical experimental permeation

procedure. A polymer membrane was secured between the two chambers of

the apparatus, after which the membrane deformed to the shape of the

o-rings used to secure its seal. The lower chamber of the cell was

placed under vacuum for approximately six to twelve hours prior to the

start of the experiment to remove any excess solvent or 02 that may have

been trapped in the film during preparation. The lower chamber was then

closed off under vacuum and an initial pressure measurement taken.






























NH(CH2CH2CN)2
---------


CH2-N (CH2CH2CN)2


[CH2-CH]I

2-(HO)C6HiCHO

HO C

CH2tlU(CH2)31N-CH 32


2-N (CH2CH2CH2NH2 2


[P]-CoSDPT reaction scheme.


CH2C1


BH 3 THF


Figure 2-4.





































sample<-

















Figure 2-5.


manometer









permeation
cell vacuum














Schematic diagram of the gas permeation experimental
set up.









Periodically, the lower chamber was monitored manometrically to

determine the extent of gas permeation and quantitatively to determine

its dioxygen ( also referred to as oxygen) content. The latter analysis

was carried out using gas samples taken from the lower chamber with a

100 ul gas tight syringe and analyzed by a Varian 3700 gas chromatograph

equipped with a thermal conductivity detector. The column employed was

an 8 foot, 1/8 inch O.D., stainless steel column containing 5A molecular

sieves heated to 30*C.

Since the gas sample taken was at a reduced pressure, it was

diluted with air while being withdrawn from the sampling port. From the

percent oxygen of the injected sample determined by the GC, the actual

percent oxygen in the lower chamber was calculated using the equation

shown below, where PL is the pressure of the lower chamber in mmHg, PH



(PL)(OL) + H L)(H) = (PH)(OD)


is atmospheric pressure, 0L is the percent oxygen in the lower chamber,

OH is the percent oxygen in the atmosphere, and OD is the percent oxygen
as determined by the GC. Using this method, the standard deviation

between samples was typically on the order of 0.1%.


Results and Discussion

The evaluation of the cobalt containing membranes is based upon a

comparison of experimental permeation data with data obtained from an

appropriate blank. Permeation data for a typical PS/[P]-CoSDPT membrane








22

are shown in table 2-1. The percent oxygen was calculated as previously

described and the percent enrichment is the difference between the

percentage of oxygen in the lower chamber and that of air, which is

assumed to be 21.0%.


Table 2-1.

Total
Time Pressure
(hr) (mmHq)

5 36.0

18 71.0

24 92.0

29.5 99.0

42 118.0


Permeation


Flux
(mmHg/hr)

7.20

3.94

3.83

3.35

2.81


data for PS/[P]-CoSDPT

Partial Pressure
Oxygen Nitrogen
(mmHg) (mmHg)

10.56 25.44

23.77 47.23

29.78 62.22

32.97 66.03

40.27 77.73


(16.4%) membrane.


Percent
Oxygen

29.3

33.5

32.4

33.3

34.1


Percent
Enrichment

8.3

12.5

11.4

12.3

13.1


According to Henry's Law, permeation of a gas


through a membrane


is a first order kinetic process. This is shown mathematically below,

where P is the permeability coefficient (cm3(STP)cm/cm2s mmHg), 1 is the

film thickness (cm), A is the film area (cm2), Vcell is the volume of

the lower chamber (cm3), (PH L) is the partial pressure differential

across the membrane (mmHg), R is the gas constant (6.24 x 104

cm3mmHg/molK), and T is the temperature (K). Assuming PH is constant



dP RTA P(PH PL)

dt 22,414Vcell 1








23

throughout the experiment, the equation can be rearranged and integrated

to give a new equation from which a plot of In(PH PL) versus time

should yield a straight line, indicating a first order process.


-RTA P
ln(PH ) = t + constant
22,414Vell 1



This analysis was carried out using the computer program detailed

in Appendix A. Using this program, the permeability coefficients for

both oxygen and nitrogen can be obtained from the slope of the lines

plotted from their respective partial pressures in table 2-1. Graphs of

these analyses are shown in figures 2-6 and 2-7. Values of 4.38 x 10-10

and 1.92 x 10-10 cm3(STP)cm/cm2s mmHg are obtained respectively.

However, in the cobalt complex containing films, it is obvious that a

slightly curved line is obtained rather than a linear one. This

indicates that the permeability coefficient of oxygen is varying with

the partial pressure of oxygen in the lower chamber. Therefore, oxygen

permeation through the membrane is deviating from a first order process

due to contributions from facilitated transport.

To better demonstrate and quantify the contributions resulting

from facilitated transport, the cobalt containing films must be compared

to an appropriate blank. Analogous nickel(II) complexes do not bind

oxygen, since they have no unpaired electrons, and are similar in

structure to the cobalt complexes resulting in membranes with similar

morphology and free space volumes. Although free volume is difficult to

reproduce in glassy polymers because of air pockets and filler



















5.1




5.0



-.J
4.9



C
4.8




4.7







Figure 2-6.


TIME (hrs.)


Plot of the natural log of oxygen partial pressure
differential as a function of time for a
PS/[P]-CoSDPT, (16.4%), membrane.



















6.5




6.4




-6.3
I



6.2




6.1








Figure 2-7.


20 30
TIME (hrs.)


Plot of the natural log of nitrogen partial pressure
differential as a function of time for a
PS/[P]-CoSDPT, (16.4%), membrane.










aggregation, the nickel complexes are the best simulation for the

unoxygenated cobalt containing films and, therefore, are ideal blanks.

Analysis of the polystyrene and nickel blanks is best made by observing


the ratio of P02

a, and describes

The a values for

reported in tabl

Certainly,

than others. An

number of films.

original film or

for a run to be


to PN2* This value is known as the separation factor,

the effectiveness of a membrane for permselectivity.

both polystyrene and nickel blank membranes are

es 2-2 and 2-3 respectively.

all films contained minor defects, some more severe

extensive amount of data was collected for a large

In several instances pinholes were present in the

leaks developed as the experiment progressed. In order

found acceptable, it was required that the first data


point have as an upper limit, a value of flux times length, ( J x 1),

below 7.0 (torr.mm/hr) at the start of the experiment. A second

requirement was that in each experiment the flux must decrease as

(PH PL) decreases i.e., as time increases. This criterion was used to
discard data sets in which leaks developed during the experiment. A

series of 11 films of polystyrene selected by these requirements and

with varying thicknesses was studied yielding a mean oxygen permeability

coefficient of 3.1 x 10-10 cm3(STP) cm/(cm2s mmHg) with a standard

deviation of +0.8 x 10-10. This permeability coefficient is

approximately one order of magnitude greater than those previously

reported.53-57 This is not surprising since polystyrene is a glassy

polymer and the method of membrane preparation will affect the transport

properties of the membrane.58'59 Systematic errors introduced by the

apparatus or sampling technique may also contribute to the deviation.










Table 2-2.


Separation data for a series of polystyrene blank
membranes.


l(mm)


0.371 + .073

0.542 + .124

0.50 + .00

0.164 + .05

0.150 + .052

0.250 + .052

0.110 + .028

0.131 + .048

0.989 + .16

1.567 + .30

0.238 + .051


J1
(torr.mm/hr)


1.7

2.3

3.4

4.3

1.5

1.2

1.0

1.9

5.4

4.3

1.5

Avg. a = 1.97


Pts.a


3F

4M

6F

4M

4M

5M

7F

9F

5F

4F

3M


a


2.52 + .02

1.37 + .02

2.14 + .02

1.40 + .01

1.72 + .01

2.19 + .04

2.27 + .02

2.34 + .03

1.97 + .05

2.76 + .20

1.03 + .16


S.D. = 0.17


a number of data points in the data set. E, M, F designate the
time span of the experiment where E is less than 12 hrs., M is
less than 29 hrs., and F is greater than or equal to 29 hrs.










Table 2-3.


Separation data for a series of nickel blank
membranes.


l(mm)


0.457 + .119

0.681 + .059

0.423 + .103


WT %


20

15

10







20

20

15

10


.12

.168

.160

.138


25 0.709 + .159


0.694

0.666

0.730


.105

.114

.079


r[P-NiSDPT

1st Cycle

J1
(torr.mm/hr)


4.8

3.1

4.0

Avg. a = 1.80

rP1-NiBr2SDPT

1st Cycle

3.9

5.2

3.0

4.4

Avg. a = 1.95

2nd Cycle

3.5

[SG1-NiSDPT

4.5

3.0

4.7

Avg. a = 1.83


Pts.a


4E

5M

5E

S.D. = 0.17





5E

5M

6M

5M

S.D. = 0.13



4M 2.



5M

5M

4M

S.D. = 0.10


a


1.79 + .01

2.05 + .02

1.57 + .02







2.26 + .03

1.80 + .12

1.94 + .06

1.79 + .08





12 + .02



1.77 + .05

1.73 + .01

2.00 + .04


a number of data points in the data set. E, M, F designate the
time span of the experiment where E is less than 12 hrs., M is
less than 29 hrs., and F is greater than or equal to 29 hrs.


0.432

0.700

0.600

0.628







29

The permeabilities were also determined from a gas mixture rather than a

single component gas penetrating through the membrane. The few studies

reported with permeations of gas mixtures through glassy polymers show

deviations in permeabilities of pure penetrants versus those of the

mixture.60 Although deviations from the literature values exist, the

important point to be considered is that all films were similarly made

and evaluated so that consistent data could be obtained and compared.

Using PN2 as an internal standard ensures that the cobalt complex-

containing membranes can be compared with their respective blanks and

the difference in oxygen partial pressures can be attributed to metal

promoted enhanced oxygen permeability and not to film preparation or

defects. To illustrate the enrichment that is obtained by facilitated

transfer, the partial pressure of oxygen in the lower chamber has been

plotted as a function of time for PS/[P]-CoSDPT (16.4% functionalized

ligand with 0.15% cobalt) membrane (figure 2-8). The change in nitrogen

partial pressure with time for this film is used as an internal standard

to calibrate the permeation experiment. The average a value from the

PS/[P]-NiSDPT blanks is used to calculate the oxygen partial pressure

corresponding to this nitrogen pressure for an identical membrane in

which facilitated transport does not occur and a first order kinetic

process is observed. Since the cobalt complex does not bind nitrogen

and facilitate its transport, this affords an ideal blank. This

calculated pressure of oxygen is representative of all diffusion

mechanisms other than facilitated transport and is also plotted in

figure 2-8. The difference between the curves represents the enrichment

obtained by a cobalt(II) facilitated transport mechanism. Again this



















E
E


LIU

U)
LU



COL
w




L-J






X
0


0 10


20 30
TIME (hrs.)


Figure 2-8.


Oxygen partial pressure as a function of time.
a) Experimental curve. b) Calculated non-facilitated
curve. c) Calculated reference curve (a = 1.00).


so
50-



40-



30



20-



10


o= EXPTL.
o = BLANK
A = AIR Po


I I 2


Po2
Po2
2


E I m










analysis was conducted using the computer program listed in appendix A.

Figure 2-8 also contains a reference curve indicating the oxygen partial

pressure which would be present if no permselectivity were achieved,

i.e. a = 1.00.

By analyzing the data using the blank nickel a values and nitrogen

permeability coefficients from the cobalt containing membranes,

systematic errors in measuring the enrichment of oxygen (such as

membrane thickness and imperfections) are cancelled out. This is

verified experimentally with the nickel blanks themselves, where a is

independent of film thickness. It was also observed in the nickel data,

that percent loading and variation in substituents on the phenyl ring

did not influence the a values within the experimental standard

deviation of + 0.2. The reaction chamber can be evacuated after 25

hours and the experiment cycled a second time with no change in a.

Table 2-4 contains the differences in partial pressures of oxygen

between the cobalt facilitated and back-calculated non-facilitated

curves (as plotted in figure 2-8) for a series of cobalt complex

containing membranes. The differences were determined graphically at

three different reaction times: when the back-calculated
B
non-facilitated partial pressure of oxygen, P02B, was 15, 20, and 25

torr. The polymer bound CoSDPT complex shows an enrichment of oxygen

over the non-facilitated process. The increasing oxygen partial

pressure at the three different reaction times can be explained by the

variance in its permeation coefficient. Early in the experiment, the

oxygen permeation coefficient is largest leading to an increase in the

difference in P02. Later in the experiment, the increased P02 value in










Table 2-4.


32

Difference in oxygen partial pressure for a series of
PS/[P]-CoSDPT membranes.


FP1-CoSDPT

1st Cycle


l(mm)


0.261 + .144

0.357 + .065

0.687 + .124


(torr.mm/hr)


2.4

2.6

4.9

Avg. P02 =

S.D. =


Difference in P02
B B
02 P02
15 20


3.7 N.A.

1.0 2.5

2.3 4.1

2.3 3.3

1.0 1.1


0.677 + .044

0.687 + .124









0.687 + .124


2nd Cycle

5.1

2.5

Avg. P02 =

S.D. =



3rd Cycle

2.9


4th Cycle

4.4


WT%


44

18.6

16.4


(torr)

P02B

25


N.A.

2.6

4.9

3.8

1.6


17.2

16.4









16.4


4.2

2.8

3.5

1.0





2.6


5.0

3.6

4.3

1.0





3.2


5.9

4.2

5.0

1.2





3.6


16.4 0.687 + .124


-2.4


-3.5 -3.9










Table 2-5.


33

Difference in oxygen partial pressure for a series of
PS/[P]-CoBr2SDPT membranes.


[P1-CoBr2SDPT

1st Cycle


l(mm)


0.528 + .072

0.671 + .047


(torr-mm/hr)


3.7

6.7

Avg. P02 =

S.D. =


Difference inBP02
P02 P02
15 20


2.9 3.9

2.6 3.5

2.8 3.7

0.2 0.3


0.640 + .051

0.747 + .074


2nd Cycle

5.8

3.7

Avg. P02 =

S.D. =


WT%


7.4

6.76


(torg)
P
P02
25


4.0

4.4

4.2

0.3


20.20

12.62


-5.8

-4.0

-4.9

1.3


-4.5

N.A.

-4.5


-5.6

N.A.

-5.6










Table 2-6.


34

Difference in oxygen partial pressure for a series of
PS/[P]-Co3FSDPT membranes.


rP1-Co3FSDPT

1st Cycle


l(mm)


0.359 + .050

0.506 + .075

0.625 + .085

0.541 + .051

Avg.


J1

(torr.mm/hr)


2.3

4.4

5.9

3.8

P02 =
S.D.=


Difference in P02
B B
P02 P02
15 20


6.0 7.1

1.0 0.6

3.0 4.0

5.4 5.0

3.8 4.2

1.3 1.6


0.359 + .050

0.541 + .051


2nd Cycle

1.4

2.4

Avg. P02 =

S.D. =


WT%


24.6

16.47

6.42

6.30


(torr)

02B

25


7.9

0.0

4.4

5.0

4.3

1.9


24.6

6.30


-2.1

0.5

-0.8

1.8


N.A.

1.0

1.0


N.A.

1.3

1.3










Table 2-7.


Difference in oxygen partial pressure for a series of
PS/[SG]-CoSDPT membranes.


[SGI-CoSDPT

1st Cycle


l(mm)


0.533
0.807
0.571
0.614
0.492


.097
.70
.082
.077
.028


0.647 + .051
0.533 + .097
0.807 + .70


(torr.mm/hr)


5.6
6.4
4.8
6.1
4.9
Avg. P02 =
S.D. =


2nd Cycle
4.2
3.2
5.6
Avg. P02 =
S.D. =


14.9
10.0
6.56
4.6
2






16.5
14.9
10.0






16.1
1.87


Difference in P02
B B
02 02
15 20


1.0
2.8
1.6
2.6
1.3
1.9
0.4



4.0
2.5
2.0
2.8
0.7



0.7
1.0
0.2
1.2


2.1
3.0
1.5
3.1
2.6
2.5
0.3



N.A.
3.2
2.2
2.7
0.7



0.0
-1.6
-0.8
1.1


WT%


Lower Surface Studies
0.384 + .037 4.6
0.358 + .062 2.7
Avg. P2 =
S.D. =


(torr)
B
P02
25


2.5
3.2
0.9
3.0
2.6
2.4
0.4



N.A.
2.5
2.5
2.5
0.0



-0.5
-1.5
-1.0
0.7









the lower chamber approaches the P1/2 value of the complex and the

facilitated transport of oxygen decreases. Eventually the cobalt

complex is fully oxygenated and does not facilitate oxygen transport.

At this point in time, no subsequent enrichment occurs, i.e. the

experimental P02 curve and blank back-calculated curves become parallel

(figure 2-8). This is observed in all cases with the different cobalt

complexes.

Second and third cycle experiments with PS/[P]-CoSDPT membranes

produced similar enrichments in comparison with the original experiment.

The fourth cycle showed a loss of enrichment presumably because of

formation of a p-peroxo complex or oxidation of the cobalt(II) complex.

While cycling experiments were not conducted on the nickel blanks with

much regularity, those that were conducted resulted in similar a values

and indicate no serious loss in separation ability. To confirm that the

observed enrichment of oxygen was due in part to the reversible binding

of oxygen and not just its desorption from the complexes already formed

during preparation of the membranes, the maximum amount of oxygen that

could possibly be bound to the cobalt centers was calculated to be

considerably less than that observed in the permeation reactions.

Facilitated transport is also demonstrated for polymer bound

CoBr2SDPT and Co3FSDPT with slightly higher enrichment values

respectively (tables 2-5 and 2-6). However, in these cases the second

cycle no longer showed any enhancement, again presumably because of

formation of a A-peroxo complex or oxidation of the cobalt(II) complex.

When CoSDPT was covalently attached to silica gel (0.5 millimoles/g) and

added to a polystyrene membrane, a smaller degree of oxygen enrichment








37

was obtained (table 2-7). Second cycles of these membranes resulted in

similar enrichments.

Attempts to achieve higher loadings of cobalt in the

functionalized polymer were moderately successful. Data reported in

tables 2-4 through 2-7 are from functionalized complexes which contain

approximately 0.30% cobalt. Using more strenuous reaction conditions

(elevated temperature and longer reaction time), loadings of 0.70% and

5.1% cobalt in P-CoSDPT were obtained. Permeation experiments and

cycling of the membranes were carried out as reported. Comparison data

are reported in table 2-8. The results indicate that with a moderate

increase in loading (0.70%), an initial increase in enrichment is

obtained; however, subsequent cycles are somewhat less, indicating

deactivation of the metal complex due primarily to increased loading.

It is not unexpected then that a loading of 5.1% had a small amount of

enrichment since approximately one out of every two ligand sites was

occupied. Expectedly, this should lead to a greater chance of metal-

metal interaction and increased formation of A-peroxodimers.

The incorporation of phenol (4% by weight) into the 5.1% cobalt

loaded membrane resulted in lowering the oxygen enrichment for the film

(P02B 15 = -1.7, Po2B 20 = -2.3, and P02 25 = -2.7 torr). It was

thought that the addition of a hydrogen bonding complex such as phenol

might stabilize the complex from deactivation. However, it appears that

it prevents reversible oxygen binding and distorts the membrane's

morphology.

The use of a more rubbery polymer, poly(butyl methacrylate), in

place of polystyrene did not result in increased facilitated transport










Table 2-8.


Comparison of the difference in oxygen partial
pressures obtained with increased cobalt loading in
PS/[P]-CoSDPT.


Difference in P02

PO02 PO2


15


2.3

3.5

2.6

-2.4



3.6

2.0

2.0



0.8


20

3.3

4.3

3.2

-3.5



4.5

2.0

2.2



1.4


(torr)

P02B
25

3.8

5.0

3.6

-3.9



5.7

2.8

2.3



N.A.


2 2.4


Cobalt
Loading

0.30%


Cycle


0.70%


5.1%


2.5 2.9









enrichment. Instead, it was observed that the experimental oxygen

partial pressure was essentially the same or slightly lower, (oxygen
enrichment at P 15 = -1.0, POB 20 = -0.5, and P B 25 = -0.25 torr),

than the back-calculated non-facilitated partial pressure. This is

explained by the fact that a blank poly(butyl methacrylate) membrane

without any additives has a high permselectivity (a = 2.2). Any

enrichment obtained with the addition of [P]-CoSDPT is overshadowed by

the increased permeability of oxygen. It is possible that this may

result in fully oxygenating the cobalt complexes at very low oxygen

pressures. This effectively blocks a facilitated transport mechanism.

In this case the complexes act as fillers and alter the permeability of

the membrane accordingly. Essentially, the rubbery polymer's diffusion

mechanism dominates that of the facilitated transport.

The mechanism of oxygen transport involved in the PS/[P]-CoSDPT

membranes described above is unlike that in liquid membranes. For

oxygen permeation using supported metal complexes in solid polymer

membranes, at least five individual processes are involved. These are

shown in figure 2-9. They include within the non-facilitated process:

solubility of the gas, diffusion of the gas, and desorption of it from

the polymer. The facilitated path also involves oxygen binding to the

metal complex, oxygen transfer to cobalt sites through the polymer, and

deoxygenation of the metal complex. On the high pressure side, the

attainment of equilibrium for oxygen solubility at the polymer interface

is an exothermic process, as is oxygen binding to the metal center.

These are not viewed as rate determining in the transport process. The

importance of the metal center in the diffusion process is much more



























SORPTION INTO POLYMER
/
/

02 EINrDING TO '.1ETAL COMPLEX



-----DIFFUSION



DESORPTION FROM POLYMER


'EOC :YGENATION OF METAL COMPLEX


Contributions to permeation mechanism.


Figure 2-9.













































02 Oz 0O


Figure 2-10.


Schematic representation of a metal facilitated
transport mechanism due to site to site interaction.







42

complex. It is proposed that the cobalt complexes transport oxygen via

a site to site transfer (figure 2-10). The oxygen bound to cobalt is in

equilibrium with oxygen that is soluble in the polymer maintaining a

reservoir of free oxygen through the film. Cobalt-oxygen complexes on

the low pressure side dissociate oxygen into the gas phase and replace

it by binding oxygen from a neighboring reservoir. This necessitates a

dispersion of the oxygen carriers throughout the membrane as opposed to

having them only on the lower surface. The latter configuration showed

no oxygen enrichment (table 2-7--Lower Surface Studies).

In conclusion, a procedure was developed for ascertaining

facilitated transfer of oxygen that uses the permeation of nitrogen in

the sample membrane as an internal standard. Using this, it has been

demonstrated that supported cobalt complexes incorporated into a

polystyrene membrane enhance the permeation of oxygen. Having

accomplished these objectives, extension of the process using higher

cobalt(II) loadings and rubbery polymers has failed to produce a more

attractive system.

The mechanism of facilitated transport in this system differs from

that in liquid membranes where diffusion of the entire complex carries

oxygen across the membrane. A site to site transfer is proposed here

and this mechanism requires that some of the cobalt be deoxygenated in

the membrane. The P1/2 value for CoSDPT in solution is reported61 to be

6.715 x 103 mmHg at 295 K. A more stable adduct forms in the

polymer,62 and its P1/2 value determines the partial pressure of oxygen

that can be attained on the lower pressure side before metal facilitated

enhancement ceases.













CHAPTER III
DEMONSTRATION OF THE FEASIBILITY AND VERSATILITY
OF MEMBRANE REACTORS


Introduction


The previous chapter discussed some of the recent technological

and commercial developments which have occurred in membrane separation

processes over the past decade. The achieved results have subsequently

spawned a new area of membrane development and application. Membranes

are no longer being used only for separations. Increasingly, they are

being extended to biochemical process applications, petrochemical

operations, and inorganic membrane reactors.1'2

Similarly, the work presented in this chapter has been sparked by

the demonstration of transition metal complexes supported in liquid and

polymer membranes for gas separations42-444651. The concept of

permselective transport of gases by such systems has led to the use of

similar membranes, or films, to catalyze reactions of gaseous

substrates. A process such as this might lead to a facile separation of

products from the catalyst. Additionally the selective removal of the

products can shift an equilibrium system.

Even though much research is being conducted within this rapidly

expanding area, the application of polymeric membrane reactors to pre-

existing commercial homogeneous processes has been widely neglected.

The objective of this research was twofold. The first was to










demonstrate the feasibility of a polymeric membrane reactor using well

characterized catalytic reactions. The second was to apply the

knowledge gained in obtaining the first goal to more reactions which

utilize the advantages of a membrane reactor. The hydroformylation of

propylene and the oxidation of 1-butene using transition metal compounds

supported in polystyrene, poly(butyl methacrylate), and composite

polystyrene/silicone gum rubber membranes were studied for demonstration

purposes. The oxidation of mustard agent simulants and the formation of

hydrogen peroxide were studied for application purposes.

The use of insoluble, polymer bound transition metal complexes to

catalyze homogeneous reactions has been reported.9-12 A polymeric

membrane reactor, however, utilizes a polymer medium for the reaction

environment. This novel method of reaction has not been explored and

thus investigation of its feasibility, illustration of its versatility,

and elucidation of the factors which influence reactivity are warranted.

The concept of using membranes for reaction stems from attempts to

combine multiple processing steps and thus achieve economically more

attractive systems. The ability to conduct a reaction within a membrane

along with the membrane's inherent ability for separation might

eliminate several purification steps which might otherwise be necessary.

To achieve this goal, layering or combining different membrane types may

provide systems which allow chemical reaction and product-reactant

separation within the same process unit. Hollow fiber membranes may

offer another attractive pathway for improving these processes. These

membranes consist of bundles of hollow fibers (10-1000 im in diameter)







45

and provide extremely high surface area per unit volume,(e.g. 500 m2 in
a 0.3 m3 reactor).
An important aspect to be considered when investigating the
feasibility of membrane reactors is how a reaction at the transition
metal center is affected by the surrounding polymer environment. As in
any discipline, the most useful information is obtained by comparing
experimental systems and results with well characterized processes. In
this case, using analogous solution chemistry as a basis for comparison
gives an accurate determination of the properties associated with
reaction in a membrane. Therefore, a brief description of the reactions
employed will be presented. A more thorough description of
hydroformylation follows in chapter 4.
Hydroformylation is the term applied to the process whereby an
alkene reacts with carbon monoxide and hydrogen to form an aldehyde.
This is shown schematically below. The reaction does not proceed in the
absence of catalysts. Typically the catalysts or catalytic precursors
employed are metal carbonyls, with cobalt and rhodium being the most
widely used. The formation of butanal from propylene is the primary
reaction of industrial significance.63

0
11
CH3CH2CH2CH


ur'H H RhH(CO)(PPh3)3
3H 2 H2, CO 1 ATH. 65C / 0

CH
CHCH
CH3CHCH3









Commercially, the straight chain isomer, n-butanal, is more

desirable than the branch chain isomer, isobutanal. This is mainly

because of the greater number of uses and applications for n-butanal.64

The hydroformylation reaction, or oxo process as it is known

industrially, is carried out under a variety of conditions. Typical

requirements when using a cobalt catalyst are total pressures of 200 to

300 atmospheres and temperatures of 120 to 1700C.65 Use of a rhodium

catalyst results in much less severe reaction conditions. Rhodium

catalysts are active at room temperature and pressure.65 Equally as

important is the selectivity of products produced. Cobalt carbonyl

catalysts will produce a straight to branch ratio up to three to one,

while modified rhodium carbonyl catalysts can produce selectivities as

high as thirty to one.65 The relatively high activity achieved under

mild reaction conditions makes rhodium an ideal catalyst for membrane

reactor demonstration.

Hydridocarbonyltristriphenylphosphine rhodium(I), RhH(CO)(PPh3)3,

was the complex chosen to catalyze the reaction in a polymer membrane.

First reported in 1968 by Wilkinson and co-workers,6671 this

homogeneous solution catalyst was shown to be active at 250C and less

than 1 atmosphere pressure. It is a precursor to the proposed active

catalytic species, RhH(CO)(PPh3)2, and has been shown to be one of the

more active precursors. It is an ideal catalyst for demonstration

purposes.

The second system chosen to demonstrate the reactor membrane

process was that of an oxidation reaction. The word oxidation can be

used to describe many things. Equally as diverse as the applications









and definitions are the products and mechanisms by which oxidation

reactions occur. The aspect of this work is limited to a small portion

of the area of oxidations and therefore the scope of material presented

will be focused only upon pertinent information. There are, however,

many reviews and references72-75 on this broad subject which are

available.

Specifically, the oxidation of terminal alkenes to their

respective methyl ketones, as shown below, was studied. This reaction

was ideal since it offered a different type of reaction scheme from that

of hydroformylation, yet was possible to achieve under mild conditions -

- a necessity for laboratory purposes. It also is a process which has

been widely studied and well characterized and thus affords an excellent

example for demonstration purposes.

0
CATALYST II
CH3CH2CH=CH2 > CH3CH2CCH3
02 1 RAT. 409C





The oxidation work presented in this chapter is based upon two

different catalyst systems which have received considerable attention

over the past twenty years. One of the first examples of selective,

non-radical alkene oxidations using molecular oxygen was demonstrated by

Dudley and Read in 1972.76 It involved a stoichiometric oxidation of a

terminal alkene in the presence of RhH(CO)(PPh3)3 or RhC1(PPh3)3 and

produced a moderately high selectivity for ketone formation relative to

aldehyde. Subsequent work indicated that an equivalent quantity of







48

triphenylphosphine oxide was produced.77 The system was made catalytic

with the incorporation of excess triphenylphosphine ligand.78 The

selectivity of the system indicates that the reaction does not proceed

by a free radical pathway. The mechanism proposed for this process is

shown in figure 3-1. It essentially involves coordination of olefin,

formation of a peroxo species, nucleophilic attack of the peroxo ligand

on the coordinated olefin to produce a five-membered peroxymetallocycle

followed by reductive elimination of the peroxymetallocycle intermediate

and its subsequent decomposition to the observed products.

Another catalytic system which was found to produce selectivities

in excess of 98% for methyl ketone formation was reported by Mimoun and

co-workers in 1978.79 It involves the use of a 2:1 ratio of

copper(II)/rhodium(III) co-catalyst. The system is unique. Not only

does it produce a high selectivity to methyl ketone, but also it

involves the incorporation of both atoms of dioxygen in ketone product

while exhibiting a zero order dependence for dioxygen. At 400C, greater

than 100 turnovers in four hours were achieved. It was demonstrated

that the incorporation of a small quantity of isopropyl alcohol

increases the activity of the reaction and results in an initial

formation of acetone as a byproduct. Typically the catalyst precursors

used were RhCl3.xH20 and Cu(N03)2.2.5H20. However many others were also

employed.

Subsequent work on this system by Drago and co-workers80-82 has

shown that in the absence of a copper(II) co-catalyst, only one oxygen

atom is incorporated into the ketone and that acetone and water are

continuously produced. It was also demonstrated, that in the absence of



































H H
H
H-



0-0


# PPh3
0--- C.~,:CH2R -

HH ""--- H


RCH2


0-CH2


PPh3


Figure 3-1.


Read and Dudley mechanism78 for the oxidation of
terminal alkenes with RhH(CO)(PPh3)3.


RCH2CH
CH2



PPh3


PPh3


PPhO

MeCOCH2R









oxygen, either hydrogen peroxide or tertiary butyl hydroperoxide could

be used as the oxidant. This was shown in both the systems with and

without copper(II) co-catalyst. Proposed mechanisms82 based upon these

observations are shown in figures 3-2 and 3-3.

The previously discussed reactions were utilized for demonstration

purposes. Based upon the strategies and successes achieved, more

advantageous systems were studied. Because of the high cost of

producing and using membranes, an advantage must be gained over existing

traditional technology in order to employ membranes in these reaction

processes. Two reactions which might provide these advantages were

tested. The first of these, the oxidation of alkyl sulfides to

sulfoxides and sulfones with oxygen is shown below. The development of

reactor membranes for this purpose is of particular importance. Their

use as self decontamination systems is being actively investigated by

the United States Army Chemical Research Development and Engineering

Center. Alkyl sulfides are simulants for mustard gas agents and new

routes to their decomposition are being investigated. Reactor membranes

which can oxidatively decompose alkyl sulfides are being studied for

their use as a polymer "self-decontaminating film" on tanks and other

weaponry.

0


CATALYST
II
(CH(CH2CH2CH2A.S

(CH3CHCHCHCS 2a Cr L s -t)
II
(CH3CH2CH2CH9)S
0











RhCI!-
LCICH3CHROH
H + R
Rh -0-C-H
CH3
H CH3CR'O
Rh
2Cu C-


2Cu


CH3CRO


2Cu -
2 Cu -


0
r--\

H-C-C-H
I I
H R


Figure 3-2.


-O0--H

CH2=CHR






Rh-O-O-K

CH2= CHR


Proposed mechanism82 for the oxidation of terminal
alkenes using a rhodium(III)/copper(II) co-catalyst.


CH2=C


H'










RhCI
-C H3CHR'OH
Rh R'
Rh -O-C-H


CH3CRb


2Cu


2Cu


CH3CRO


R
Rh O-C-H
CH3
CH3CRO0


Rh


-O-H

CH2=CHR
Cu'


R


R-C-CH3
O
Rh 0


HR


Figure 3-3.


Rh-O-O-C-H
CH3


CH2"CHR


Rh O-O-C-H
CH2=CHR CH3


Alternative mechanism82 for the oxidation of terminal
alkenes using a rhodium(III)/copper(II) co-catalyst.







53
The selective oxidation of sulfides to sulfoxides using transition

metal catalysts has been the object of an extensive quantity of

research.83-86 Most of this reported work utilizes alkyl

hydroperoxides8789 or hydrogen peroxide9093 as the oxidant and results

in the formation of sulfones as a byproduct. The use of 02 or air with

these transition metal catalysts, in most cases, results in similar

product formation.9496 Ledlie and co-workers used RuC13.xH20 to

catalyze the oxidation of butyl sulfide.97 Later, Riley utilized

ruthenium(II) dimethyl sulfoxide catalysts in similar systems and

proposed two possible mechanisms whereby a ruthenium(II)/ruthenium(IV)

cycle operated.98-00 These are shown in the equations la and b, and 2.


"Ru(II)" + 02 <=====> "Ru(IV)" + 022 (la)

SR2 + H202 + ROH <=====> S(O)R2 + H20 + ROH (Ib)


"Ru(IV)-SR2" + H20 <=====> "Ru(II)" + S(O)R2 + 2H+ (2)


The demonstration of ruthenium-oxo complexes in the catalysis of a

variety of organic substrate oxidations101-104 indicated that such

catalysts might be suitable sulfide oxidation catalysts. Subsequently,

Drago and co-workers have shown both cis-dioxo bis(2,9-dimethyl-l,10-

phenanthroline)-ruthenium(VI)hexafluorophosphate, [Ru(0)2(dmp)2](PF6)2,

and the trimer trisaquo-oxo-hexapropianato-trisruthenium(III)

propionate, [Ru30(prop)6(H20)3](prop), to be capable of oxidizing butyl

sulfide to butyl sulfoxide05-107 in acetonitrile under relatively mild







54
reaction conditions of 1000C and 2 atmospheres of oxygen. The catalysts

are shown in figure 3-4a and b respectively.

The second application of membrane reactors is focused on the

formation of hydrogen peroxide. The commercial production of hydrogen

peroxide is accomplished by the anthraquinone autoxidation processl8 or

by direct combination of hydrogen and oxygen.109 Presently, there is a

rapidly expanding market for hydrogen peroxide. It was estimated that

365 million pounds of H202 were used in 1987 for many different

applications. The largest of these included chemical production, pulp

and paper processes, and environmental purposes. It is estimated that

in 1992 590 million pounds will be utilized per year.110 In order to

keep up with such a rapidly expanding need, many companies are being

forced to expand and modernize their existing plants or to develop new

methods for H202 production.109,110

The anthraquinone autoxidation process was first commercially

operated in Germany during World War II. While many new processes have

been developed, all retain the basic features of the Riedl-Pfleiderer

process0 shown in figure 3-5. The process involves catalytically

reducing an anthraquinone to its corresponding anthraquinol which is

then oxidized in air to reform the anthraquinone and produce hydrogen

peroxide. The process has remained essentially the same for almost

fifty years. Little improvement has been achieved in the numerous

separation and purification steps which are required to maintain an

active system.

The hydrogenation portion of the reaction requires a chemically

stable, water insoluble, nontoxic solvent with a high flash point













































Figure 3-4.


(PF6)2


Sulfide oxidation catalysts: a) [Ru(0)2(dmp)2](PF6)2
and b) [Ru30(prop)6(H20)3]+


























0
NCH2CH3


0


OH
r N N CH2CH3


OH


PRLLADIUM BLACK
H2 60C






0,2
H20
------- >


OH
NCH2CH3


OH


CH3


+ H202


Anthraquinone autoxidation process.


Figure 3-5.









and low volatility. Typically, a 50:50 mixture of benzene and C7 C9

secondary alcohols is used. One of the major problems associated with

this system involves maintaining a soluble working material. Reduction

is carried out using slightly elevated partial pressures of hydrogen at

temperatures under 1000C and conversion is limited to 50% to minimize

anthraquinone-deactivating secondary reactions. Catalysts which have

been utilized for this process step include supported Raney nickel and

palladium.108

The oxidation of the resulting anthrahydroquinol is easily

accomplished without the aid of a catalyst. The rate is dependent upon

the pressure of oxygen and temperature of the reaction. The most

critical portion of this step is in the removal of the reducing catalyst

and the subsequent phase boundary surface area between the organic

solvent and water used for H202 extraction. Regeneration of the working

solution is required to remove all inactive anthraquinone derivatives

formed as well as any aqueous solution retained in the system.

A second method used for hydrogen peroxide production could be

considered to be in its infancy by comparison with the anthraquinone

process. Only within the past ten years has the ability to directly

combine hydrogen and oxygen to produce hydrogen peroxide become

available.109 While little has been reported of its mechanistic detail,

the procedure involved is rather straightforward. The reaction is run

in an aqueous medium using palladium on adsorbent carbon as a

hydrogenation catalyst with superatmospheric pressures (200 to 4000

psig) of hydrogen and oxygen. The reaction is conducted under mild

temperatures (0 to 500C) and produces an aqueous solution which is









approximately 20% H202 by weight. The use of an all-aqueous reaction

medium provides a safer reaction process since it does not produce

mixtures of H202 and an organic component. Application of these

processes to membrane reactors may similarly produce a system which

reduces the number of process steps required to maintain an active

system while operating under safer reaction conditions.

The primary goal of this project was to demonstrate the ability to

achieve reaction within a polymer environment and, based upon the

achieved success, apply the process to reaction systems which would be

advantageous for one reason or another to carry out within a membrane.

The work presented supports this objective. The process developed is an

extension of pre-existing catalyst technology and is not an attempt at

discovering new catalysts.



Experimental


The complexes studied and/or used in this work were prepared from

reagents which were either purchased or previously synthesized. The

following procedures can be used to prepare these compounds. The

solvents used were purified by storage over 4A molecular sieves under an

atmosphere of nitrogen. Characterization of the complexes was conducted

using a Nicolet 5DXB FTIR Spectrometer. Identification and

quantification of reaction samples were conducted using a Varian Model

3700 gas chromatograph, a Hewlett-Packard 5890A gas chromatograph in

connection with a Nicolet 5DXB FTIR Spectrometer and a Varian Model 3400

gas chromatograph in connection with a Finnigan Mat Model 700 Ion Trap









Detector. Elemental analyses were conducted by the University of
Florida Department of Chemistry Microanalysis Service.
Hydridocarbonvltris(tripheny1phosphine)rhodium(I), RhH(CO)(PPh313_
Hydridocarbonyltris(triphenylphosphine)rhodium(I), RhH(CO)(PPh3)3,
was prepared according to the literature.111 Experimental analysis:
%C = 71.8, %H = 5.12, %N = 0.00. Theoretical analysis: %C = 71.9,
%H = 5.05, %N = 0.00. Melting point 120 1220C (lit. 121-1220C). Ir
analysis (nujol mull): Rh-H: 2041 and 785 cm-1 (lit. 2040 and 786 cm-1),
C=0: 1918 cm-1 (lit. 1923 cm-1).
Trifluoroacetatotris(triphenvlphosphine)rhodium(I), Rh(02CCF3)(PPh3)3
Trifluoroacetatotris(triphenylphosphine)rhodium(I),
Rh(02CCF3)(PPh3)3, was obtained from Dr. Cindy Getty and Steve
Showalter. It can be prepared11213 from Rh2(02CCF3)4 according to a
method similar to that reported by Drago and Telser.114
Dioxo-bis(2,9-dimethyl-1,10-phenanthroline)ruthenium(VI) hexafluoro-
phosphate, [Ru(0)2(dmp)121PF612

[Ru(0)2(dmp)2](PF6)2 was obtained from Dr. Cindy Bailey. It can
be prepared with moderate difficulty from the oxidation of
[Ru(dmp)2(H20)2]2+ with ceric ammonium nitrate, (NH4)2Ce(N03)6.101115
Trisaquohexapropionatotrisruthenium(III) propionate,

rRu30(Drop)6g(H2031(prop)
The ruthenium trimer [Ru30(prop)6(H20)3](prop) was obtained from
Dr. Shannon Davis. It can be prepared from RuC13.xH20.116
Membrane Preparation
The membranes employed in this study were prepared using a casting
technique. The appropriate polymer was dissolved in a suitable solvent,







60
most often toluene or methylene chloride. The transition metal complex

to be used for a specific reaction was then dissolved in the same

solvent and added to the viscous polymer solution. The resulting

solution was then transferred to an aluminum foil form. Evaporation of

the solvent at room temperature resulted in the formation of uniform

polymer membranes with an average thickness between 0.5 and 1.0 mm.

These were used without further treatment.

Permeation Apparatus

The apparatus used for the membrane reactions, (shown in

figure 3-6), was designed to provide support of the membrane under

experimental reaction conditions. The film is supported both on the low

and high pressure sides by stainless steel metal frits (51# pore size).

Temperature of the reaction is controlled by a Variac and thermal tape

in conjunction with a temperature controller.

Permeation Procedure

A typical permeation experiment is described in the following

paragraph. The membrane to be used is secured between the two chambers

of the reactor using o-rings and the two parts screwed together to

ensure a proper seal. Both chambers are evacuated for approximately 1

hour to remove any excess solvent trapped in the membrane. The lower

and upper chambers are then closed off under vacuum and the upper

chamber filled with one atmosphere of the reactants. The gases permeate

through the membrane during which time reaction occurs. The products

and unreacted substrates then desorb from the film into the evacuated

chamber. Periodically, both chambers are monitored to determine the

extent of reaction. This is done using FID gas chromatography using a








61










Needle Volve
l;'i vi //// / / ///-i /li


Polymer "
Membrane---
-. .. Cd'vJ.. ^.. .


~J3 "Rubber O-rings


Needle Volve
I


Sintered S.S. Plale
0-12 cm Ihick
5 micron pore size


Reactor membrane apparatus.


'


Figure 3-6.









2.5m stainless steel column packed with either porapak Q or

diethyleneglycol adipate. GC-FTIR, and GC-MS were also run to confirm

all of the product identities.


Results and Discussion


Hydroformylation of propylene

The hydroformylation of propylene was the first reaction attempted

in order to demonstrate the reactor membrane process. The reaction was

carried out at 650C using a 1:1:1 gas mixture of the reactants hydrogen,

carbon monoxide, and propylene at a total pressure of 1 atmosphere with

a polystyrene membrane which contained 2.14 x 10-4 moles of

RhH(CO)(PPh3)3. The pressures in the two chambers equilibrated over a

period of three days. Gas samples were taken periodically, however with

a decreasing pressure gradient, it is impossible to report the data on

an activity curve as would normally be done with a catalytic run.

Alternatively, the amount of product produced within a given period of

time is reported. In this case, 4.7 x 10-10 moles of butanal and

6.6 x 10-10 moles of isobutanal was produced in 72 hours. While it is

true that this is an extremely small quantity of product and represents

a very small amount of conversion, it is, never-the-less important since

it is the first demonstration of reaction within a polymer environment.

There are many factors which might account for the low

conversion. First, it must be considered that the reactants are gases

and therefore their concentrations in the membrane are appreciably less

than those of an analogous solution. The interaction of substrate and









catalyst is reduced because of this lower concentration of reactants.

It is entirely possible that within the polymer, there is little

catalyst substrate interaction occurring as a result of the reactants

permeating through the film relatively rapidly. This is heightened by

the slow diffusion of reactants within the film and the immobility of

the catalyst. It is likely that a large fraction of the reactants pass

through the film via pores and channels without ever coming into contact

with the catalyst. Coupled with this problem is the fact that three

different reactants must make contact with the catalyst for reaction to

occur.

Considering this, a possible way to improve upon the extent of

reaction would be to alter the diffusion and solubility characteristics

of the gases in the polymer. To accomplish this, a different polymer

was chosen for the reaction medium. Poly(butyl methacrylate), PBM, was

the polymer chosen to demonstrate how the diffusion and solubility

characteristics of the polymer could influence the reaction

characteristics. Pure PBM has a glass transition temperature (Tg) of

27C whereas the corresponding value for polystyrene is 1000C. PBM is a

rubbery polymer under reaction conditions and its diffusion and

solubility properties are considerable different than of polystyrene,

PS, a glassy polymer under the same reaction conditions.

Using reaction conditions identical with the previously mentioned

reaction, in conjunction with 3.9 x 10-5 moles of catalyst contained in

the PBM film, 1.1 x 10-8 moles of butanal was produced with only trace

quantities of isobutanal present. This is a remarkable observation

since it demonstrates just how great an effect the polymer medium can







64
have on the reaction. This result can be likened to running a solution

reaction in which two entirely different solvents are utilized. Perhaps

the most important observation is not in the increased activity, but

rather in the drastic improvement in selectivity that is achieved.

The use of a different polymer shows that the extent of reaction

can be altered by changing the diffusion and solubility properties of

the polymer. It was thought that a greater extent of reaction might be

obtained through optimization of the reaction conditions. This can be

accomplished through two means. The first involves the addition of

excess ligand. The second pertains to increasing the reaction

temperature and pressure. This is carried out using 2.4 x 10-6 moles of

RhH(CO)(PPh3)3 in conjunction with five times as much triphenylphosphine

in a PBM film at 1000C. A pressure of two atmospheres of synthesis gas

(H2 and CO), in a 1:1 ratio, is placed on both sides of the membrane and

left for a period of two days. At the start of the reaction, 2.0 mL of

propylene was injected into the upper chamber creating a small pressure

differential (50 mmHg). This experimental setup ensures complete

saturation of the membrane with synthesis gas and the partial pressure

of propylene is used to facilitate its permeation through the membrane.

Within twenty four hours, 2.2 x 10-7 moles of butanal are produced with

trace amounts of isobutanal also observed.

The best results are achieved using a system that contains

Rh(O2CCF3)(PPh3)3 in PBM with a fivefold excess of ligand under the same

reaction conditions reported previously. It produces 8.1 x 10-7 moles

of butanal and trace amounts of isobutanal. This catalyst had

previously been studied112 and shown to be efficient for the









hydroformylation of liquid substrates in solution. In this reported

case, a five molar excess of triphenylphosphine in neat 1-hexene

produced 225 turnovers and a 4.3 to 1 ratio of straight to branch chain

aldehyde. Comparison of the polymer and solution reaction mediums

reveals that they have similar dielectric constants. A summary of the

membrane reactions is shown below in table 3-1.

Oxidation of 1-butene

The second system studied for demonstration purposes resulted from

a serendipitous discovery. When air accidently leaked into the initial

hydroformylation system described above, the rapid production of acetone

from propylene was observed. This product formation appeared to occur at

a much faster rate than that of the hydroformylation products. This

discovery led to attempting the oxidation of terminal alkenes to their

respective methyl ketones for the demonstration of reactor membranes.

Again the first membrane studied is a PS film that contains

3.9 x 105 moles of RhH(CO)(PPh3)3. Reaction conditions employed were

slightly milder than those of the hydroformylation systems. Temperature

is maintained at 400C and a 1 atmosphere mixture of 02 and 1-butene is

charged into the upper chamber. After three days, 7.9 x 10-9 moles of

2-butanone is produced selectively. Data for this reaction and others

similar in nature are reported in table 3-2. The first thing observed

with this demonstration reaction is that under milder conditions, a

substantially larger quantity of product is produced relative to the

similar hydroformylation system. This indicates, as expected, that

different types of reactions are influenced to varying extents by the

surrounding environment.










Hydroformylation of propylene with RhH(CO)(PPh3)3.


Moles
Butanal

4.7x10-10


Turnovers
Butanal

2.2x10-6


1.1x10-8 2.8x10-4

2.2x10-7 9.2x10-2

8.1x10-7 3.4x10-1


Moles
Isobutanal

6.6x10-10


Turnovers
Isobutanal

3.1x10-6


trace

trace

trace


a turnover is moles of product/mole of catalyst.

b reaction conditions are 65C, 1 atm. H2, CO, and propylene
(1:1:1 mix) in the upper chamber, and reaction time is three days.

c membrane contains 1.2x10-5 moles triphenylphosphine at 1000C.
Both chambers contain 1 atm. of H2, CO (1:1 mix). Upper chamber
contains an additional 2.0 mL of propylene (1 atm. pressure).
Reaction time is one day.

d Rh(02CCF3)(PPh )3 catalyst in place of RhH(CO)(PPh3)3 under
identical conditions to note c. Reaction time is one day.


Membrane

PSb

PBMb

PBMc

PBMd


Moles
Catalyst

2.14x10-4

3.9 x10-5

2.4 x10-6

2.4 x10-6


Table 3-1.









Oxidation of 1-butene.


Catalyst

RhH(CO)(PPh3)3

RhH(CO)(PPh3)3

RhH(CO)(PPh3)3

RhH(CO)(PPh3)3

RhH(CO)(PPh3)3

Rh(III)/Cu(II)b

Rh(III)/Cu(II)b

RhH(CO)(PPh3)3


Conditions


2nd cycle


2nd cycle

600C


600C

1-hexene


Moles
2-butangne
( x 10-)

7.9

14.

12.

47.

24.

200.

160.

240.


Turnovers
2-butanone
( x 10')
2.0

3.6

3.1

12.

6.2

25.

20.

62.


a standard reaction conditions: 3.9x10-5 moles catalyst at 400C
with 1 atm. 02 and 1-butene (1:1 mix). Reaction time is three
days.

b 8.0x10-5 moles RhCl3.xH20 and 1.6x10-4 moles Cu(N03)2.2.5H20.


Membrane

PS

PS

PBM

PBM

PBM

SGR/PS

PBM

PBM


Table 3-2.







68

The same quantity of catalyst as was used in the previous reaction

was supported in a PBM film. Under similar reaction conditions, an

increase in the amount of product was observed. In this case,

1.4 x 10-8 moles of 2-butanone was produced. Cycling experiments were

conducted by reevacuating each system and refilling the upper chambers

with reactants. Essentially all membranes are saturated with reactants

at the start of the second cycle. This results in almost a twofold

increase in the production of 2-butanone for the polystyrene system and

nearly a fourfold increase with poly(butyl methacrylate). Again this is

best explained by the theory that saturation of the membranes with

reactants has occurred resulting in increased catalyst-substrate

interaction. Improved reactivity is the end result of this increased

interaction. Raising the reaction temperature to 600C while using the

PBM/RhH(CO)(PPh3)3 membrane also improves reactivity in the butene

system. This is expected for a temperature dependent reaction.

Another reaction using the liquid substrate 1-hexene was conducted

to verify that higher substrate concentration within the membrane would

indeed result in increased reactivity. Five drops of the substrate was

placed on the film and the reaction run with the same quantity of

catalyst and reaction conditions. In three days, 2.40 x 10-7 moles of

2-hexanone are produced selectively. This lends support to the idea

that the reactivity of the membrane can be increased dramatically under

more solution-like reaction conditions.

Based upon the observed results, a mechanistically different

reaction was attempted. It involved the use of a Rh(III)/Cu(II) co-

catalyst system in the oxidation of terminal alkenes. A silicone gum









rubber polymer containing 8.0 x 10-5 moles of RhC13.xH20 and 1.6 x 10-4

moles of Cu(NO3)2*2.5H20 along with a approximately 1.0 ml of

isopropanol was layered on a polystyrene blank membrane. This resulted

in the production of 2.0 x 10-7 moles of 2-butanone in 72 hours. The

same catalyst system contained in a PBM film at 600C was not quite as

reactive (1.60 x 10-7 moles in 72 hours). These observations show that

mechanistically different reactions such as this can be very active.

This supports the conclusion that the reaction type itself plays a large

part in determining the extent of reaction within the polymer framework.

Oxidation of butyl sulfide

Based upon the successes achieved in the above reactions,

attention next turned towards processes in which membrane use may be

more economically attractive. The catalytic oxidation of the simulant

butyl sulfide is studied with (1) Ru(Cl)2(PPh3)3, (2)

[Ru(0)2(dmp)2](PF6)2, and (3) [Ru30(prop)6(H20)3](prop) in

polystyrene/silicone gum rubber composite membranes. Composite catalyst

membranes are obtained by evaporating a catalyst/silicone gum

rubber/toluene solution onto a preformed blank polystyrene film. The

ruthenium membrane oxidations of butyl sulfide are carried out by

coating the upper side (the side exposed to the higher pressure) of the

films with a small amount of the liquid substrate. The top chamber of

the reactor is then filled with 30 psi 02 and the bottom chamber is

closed off under vacuum. Temperature of the reaction is maintained at

900C. The pressures within the two chambers equilibrate over a period

of two days. The gas which collects in the bottom chamber is determined

to contain large percentages of butyl sulfoxide, butyraldehyde, and








70
propionaldehyde, along with a small quantity of unreacted butyl sulfide.

Trace quantities of 1-butene and 1,2-epoxybutane are also observed using

GC-MS.

In all membrane reactions, the total amount of products formed

appears to be substantially smaller in comparison with solution data. A

better method of analysis in this particular case would be to compare

the percentage conversion of butyl sulfide. In analyzing these membrane

reactions, the percent conversion is based upon the amount of butyl

sulfide and products which have passed through the membrane and not on

the original amount exposed to the film. This is the most relevant mode

of analysis since the employment of these materials as self

decontaminating films focuses only on the gases which permeate the film

and their relative amounts.

Table 3-3 lists the mole percent conversion for each of the

composite membrane catalysts systems. It should be noted that a

modified reactor design employing a smaller film with less catalyst is

used in the phenanthroline derivative system. In light of this, a

better judge of the membrane activities is made by comparing turnovers

of the respective catalyst membranes. A turnover is defined as the

number of moles of product produced per mole of catalyst. This gives an

accurate representation of the reaction assuming that an excess amount

of substrate was exposed to each film. Turnover data is presented in

table 3-4 and indicates that the ruthenium trimer is the best overall

catalyst employed in the membranes. Evidence which supports the th ry

that a large quantity of substrate may never come in contact with the









Table 3-3.


Butyl sulfide membrane conversion data for various
ruthenium catalyzed oxidations by 02 at 1 atm. and
90 C.


Mole Percent


Reaction
Time (hrs)


Butyl
Sulfide


Butyl
Sulfoxide


Butanal Propanal


Total
Conversion


Ru(C1)2(PPh3)3


11.3

15.5


40.0

40.0


[Ru30(prop)6(H20)3](prop)


2.0

0.3


[Ru(02)2(dmp)2](PF6)2


85.6


13.2


14.4


48 78.0


45.9

42.6


97.2

98.1


10.8


85.8

95.9


89.2

96.4


2.6 19.4


trace 22.0









Butyl sulfide membrane turnover data.


Catalyst

Ru(C1)2(PPh3)3

[Ru30(prop)6(H20)3](prop)

[Ru(0)2(dmp)2](PF6)2


Butyl
Sulfoxide

0.27

1.00

0.40


Turnovers ( x 102)


Butanal

0.97

1.49

4.68


Propanal

1.11

63.70


Turnover data are reported at a reaction time of 24 hours based on
data in table 3-3.


Table 3-4.


Total

2.35

66.19

5.08









catalyst in the membrane is obtained by comparing tables 3-3 and 3-4.

It is apparent that a larger conversion percentage can be obtained by

increasing the catalyst loading in the membrane and thus increasing the

extent of interaction between substrate and catalyst.

Comparison of the membrane data can be made to their solution

analogues0 as well. Reported mole percent conversions of butyl

sulfide for the three ruthenium catalysts in solution under similar

reaction conditions is presented in table 3-5. The relatively high

percentage of conversion compared to solution studies indicates that

reactor membranes may prove to be useful as self decontamination

systems. The data obtained is very encouraging and indicates that the

catalyst is active when incorporated into a polymer network. Though the

activity of the membranes is lower than the solution systems, based on

the amount converted per unit time, the high conversion of material

passing through film is a desired result.

Formation of hydrogen peroxide

The production of hydrogen peroxide has been accomplished by the

anthraquinone autoxidation process and by direct combination of hydrogen

and oxygen. The application of these two systems to membrane reactors

is yet another process which has shown membrane reactors to be a

versatile process. One part of this research incorporates the basic

reaction scheme outlined in the introduction. As noted earlier, a major

drawback associated with the anthraquinone autoxidation process is

maintaining a soluble system. Several reaction and separation steps are

needed to maintain an active process. This work represents a novel










Table 3-5.


Butyl sulfide solution reaction conversion data for
various ruthenium catalyzed oxidations by 02 at
900C.


Mole Percent


Reaction
Time (hrs)


Butyl Butyl
Sulfide Sulfoxide


Total
Butanal Propanal Conversion


Ru(C1)2(PPh3)3


50.3

59.1


[Ru30(prop)6(H20)3](prop)

2.6

22.8


40.7


[Ru(02)2(dmp)2](PF6)2
-- 0.7


25.9

28.1


2.0

2.8


Source of data is reference 106.


50.3

59.1


49.7

40.9


97.4

77.2

59.7


99.3

72.1

69.1


22.8

41.3


27.9

30.9







75

approach aimed at minimizing the number of required process steps while

maintaining the solubility and stability of the working material in both

its oxidized and reduced states. This would result in a more

economically attractive system. The reaction is conducted by

incorporating the catalytic system into a rubbery polymer membrane, PBM,

which is used as the medium for reaction. The second system under

investigation, the direct combination of oxygen and hydrogen, is

performed by incorporating a palladium catalyst into a PBM membrane and

using it as the reaction medium.

Flat reactor membranes are formed from poly(butyl methacrylate)

using a casting technique. Depending upon the system being studied,

either palladium black or a mixture containing 2-ethyl anthraquinone and

palladium black are incorporated in the polymer membrane. A different

apparatus than was used in the previous systems is used to support the

reaction. It is shown in Figure 3-8 and consists of two glass chambers.

The membrane is sealed between them using an 0-ring joint and clamp.

The reaction is facilitated by passing hydrogen gas over the bottom side

of the membrane and by saturating an aqueous solution on the top side of

the membrane with oxygen. Reaction occurs as the two gases diffuse into

the membrane and come in contact with the catalyst system. The

temperature of reaction is maintained by placing the reactor in an oil

bath at the desired temperature. A slightly positive pressure of

hydrogen is maintained in the lower chamber. Reaction occurs over a

four day period. The resulting hydrogen peroxide is extracted into the

aqueous layer and titrated with potassium permanganate to determine the

extent of reaction.




















OXYGEN IN


HYDROGEN IN

I


MEMBRANE


OXYGEN OUT


AQUEOUS SOLUTION
HYDROGEN OUT


0-RING JOINT


Hydrogen peroxide membrane reactor.


Figure 3-7.









There are several key variables in this process that have been

investigated. Included among these are temperature dependence of the

process, the effect of added promoters, and the effect of cycling the

membranes. Data for the performed reactions are contained in table 3-6.

The results indicate that the anthraquinone systems are the most active.

Increased temperature also results in a more active system. Reaction

using only palladium black was accomplished with comparable activity to

the anthraquinone system. The best system for direct combination of

hydrogen and oxygen is one which incorporates an acid and halide

promoter in the film and employs higher reaction temperatures. Attempts

to find a more active system using a palladium colloid suspended in a

polymer film failed when the poly(vinyl alcohol vinyl acetate) membrane

used in the reaction dissolved during reaction.

In conclusion, the comparison of all membrane reactions with their

respective homogeneous analogues results in some notable distinctions.

The membrane reactions appear to be substantially less reactive. This

is due largely in part to the relative amount of time that the substrate

spends in contact with the catalyst. A possible way of improving upon

this would be to increase catalyst concentration. A thicker membrane

might also provide for better interaction. Comparatively, a solution

reaction occurs with the substrate continuously diffusing in and out of

solution. This results in an infinite amount of exposure with the

catalyst. In a membrane reaction, the substrate is contained in the

film for a much smaller amount of time, effectively limiting the extent

of reaction. Another observation that would explain the difference in

the reactivities of membrane and solution reactions is the relative









Hydrogen peroxide formation.


CATALYST


PROMOTER


Pd BLACK & NONE
2-ETHYL ANTHRAQUINONE


REACTION
CONDITIONS

600C


1

2.2


CYCLES (moles H202
2 3


Pd BLACK & NONE
2-ETHYL ANTHRAQUINONE

Pd BLACK NONE

PD BLACK H+, Br-

Pd BLACK NAFION, H+, Br-

Pd BLACK NONE

Pd BLACK H+, Br-

Pd BLACK NAFION, H+, Br-


950C


60C

600C

600C

85C

850C

850C


0.9


Reaction conditions: PBM film layered on PS, 20.0 mL aqueous
solution for f ur day cycles. [H+] = 1 M HC1, [Br ] = 1.3 x 10- M
KBr, 3.6 x 10- moles Pd black, and/or 1.5 x 10-4 moles 2-ethyl
anthraquinone.


x 106)
4

1.2


Table 3-6.









difference in substrate concentrations contained in each reaction

medium. This directly affects the number of catalyst substrate

interactions which occur.

The initial objective of this project has been obtained.

Demonstration of the feasibility and versatility of transition metal

reactor membranes has been achieved with both hydroformylation and

oxidation reactions. The activities of the membranes are low as a

result of the engineering design of our experiment. The use of hollow

fiber membranes may lead to a more productive system. For example, the

best hydroformylation system studied produces 8.1 x 10-7 moles of

butanal in 24 hours or 4.06 x 108g of butanal per minute using a film

with a surface area of 5.0 x 10-4 2. A typical hollow fiber membrane

unit using 100pm fibers contained in a 0.3 m3 reactor has a surface area

of 500 m2. Transposing our system to this reactor would produce 4.0 x

10-2g of butanal per minute. This may be one way of increasing

production capabilities in membrane reactors. Optimization of reaction

conditions has been shown to result in remarkable increases in

reactivity.













CHAPTER IV
HETEROGENEOUS HYDROFORMYLATION OF PROPYLENE USING SUPPORTED
RHODIUM CATALYSTS IN A CONTINUOUS GAS FLOW REACTOR


Introduction


The previous chapter demonstrates the feasibility and versatility

of facilitating chemical reactions within an organic polymer

environment. One remarkable attribute of membrane reactions lies in

their ability to achieve high selectivity for a particular product.

This is evidenced in the hydroformylation of propylene to n-butanal with

a high degree of selectivity. A major drawback that is associated with

these membranes is in their extremely low activity. The work presented

in this chapter involves the development of a gas phase, heterogeneous

catalyst based system resulting in substantially increased activity

while maintaining the selectivity associated with the polymer membrane

reactors.

The hydroformylation of propylene is the reaction chosen to

demonstrate this objective. The ability to use a gaseous substrate, the

possibility of producing either straight or branched chain product, and

the availability of a catalyst which is active under moderate reaction

conditions makes it ideal for study. The reaction is previously

described in chapter 3. A more detailed discussion of the reaction is

presented in this section.









Hydroformylation is the process whereby an alkene reacts with

carbon monoxide, CO, and hydrogen, H2, to form an aldehyde. The

hydroformylation of propylene results in the formation of both n-butanal

and isobutanal as shown below.


0
0 \CH

CH3CH=CH2 RHCO(PH CH3CH2CH2CH + CH3CHCH3
CHCH=C2 H2, o 1 ATm. 65*0C



The reaction does not proceed in the absence of a catalyst.

Commercially, two transition metals, cobalt and rhodium, are used to

form active catalyst complexes and are usually employed as the metal

carbonyl complexes. First discovered by Otto Roelen in 1938,117 the

process has since grown to immense proportion. In 1980, over eight

billion pounds of aldehydes or their derivatives were produced by the

process, with the formation of butanal from propylene being the primary

reaction scheme.63

Commercially, the straight chain isomer, n-butanal, is more

desirable than the branched chain isomer, isobutanal. This is because

of the greater number of uses and applications for n-butanal.64 As

shown below, n-butanal can be converted through a variety of steps to

n-butanol and 2-ethylhexanol. The latter can be converted into

plasticizers for polyvinylchloride resins.

The hydroformylation reaction, or oxo process as it is known

industrially, can be achieved using a variety of conditions. Typical

requirements for a cobalt catalyst system require total pressures of 200











SCHO ,C02H
CH3CH2CH2CHO _H0 CH(CH2)2CH'C CH3(CH2)3CH
-H2 C25 CH2CH3
cat. H2 cat. H2
CHO H2
CH3CH2CH2CH20H CH3(CH2 3CH CH
HH2CH3( cat.
CH2CH3
,CH2OH
CH3(CH2)3CH
NCH2CH3




to 300 atmospheres and temperatures of 120 to 1700C.65 Use of rhodium

catalysts result in much less severe reaction conditions. Activity can

be observed using room temperature and pressure.65 Equally as important

is the selectivity. Cobalt carbonyl catalysts produce a straight to

branch ratio up to three to one, while modified rhodium carbonyl

catalysts produce selectivities as high as thirty to one.64 The

selectivity obtained is dependent upon a number of factors including the

ligand environment of the metal, the presence of excess ligand in the

system, and the pressure of gaseous reactants used. The relatively high

activity achieved under mild reaction conditions makes rhodium an ideal

catalyst for continuous gas flow reactor demonstration.

Modified rhodium oxo catalysts were introduced into commercial use

in 1976.118-120 One of these, hydridocarbonyltristriphenylphosphine

rhodium(I), RhH(CO)(PPh3)3, was the catalyst chosen to catalyze the

reaction. First reported in 1968 by Wilkinson and associates,66-71 this

homogeneous catalyst was shown to be active at 25C and less than 1

atmosphere pressure. It is a precursor to the proposed active catalytic







83

species, RhH(CO)(PPh3)2, and has been demonstrated to be one of the more

active rhodium precursors. It does not react to form inactive metal

clusters like many other rhodium carbonyl complexes.

Selectivity is controlled by a number of factors. The use of

polar solvents as the reaction medium leads to increased reactivity and

produces increased selectivity.121 Most importantly, the addition of

excess triphenylphosphine results in increased linear selectivity. The

reaction has even been carried out in molten triphenylphosphine.64 A

disadvantage though, is that the rate of hydroformylation decreases with

the addition of excess phosphine. The increase in selectivity has been

attributed to increased steric hinderance while the decreased activity

can be explained by the effect of shifting the equilibrium from a

proposed active catalytic species towards the precursor. This

equilibrium is shown in the equation below.


RhH(CO)(PPh3)3 <==========> RhH(CO)(PPh3)2 + PPh3


The widely accepted mechanism of this reaction is depicted in

figure 4-1. As shown, the reaction can occur by either a dissociative

or associative pathway with the first step being the subsequent

dissociation of a triphenylphosphine ligand.65 Coordination of carbon

monoxide is the next proposed step. Next, olefin coordination occurs.

It is the most critical step. Subsequent addition of the olefin to the

rhodium hydride bond determines whether linear or branched chain

aldehyde is formed. Markonikov addition results in the formation of a

branched alkyl complex, while anti-Markonikov addition results in the




















H H
-L
'-. I -L I
Rh-CO L-Rh-CO
CO +L
CO CO


Coll


H
II
L-Rh--L

O


H
L. I
L>I
OC
CO



11
R


L--Rh--CO

CO


f-RCH 2CHCHO
R R


0 CO
H. H rds
RRh *. L-Rh--L
oc" "L
CO


Figure 4-1.


/ 11+L

R


Rh--CO
L A
CO


Associative and dissociative hydroformylation
reaction mechanism.


~cp









formation of a linear alkyl complex. Carbon monoxide migration and

insertion to form the acyl complex is the next proposed mechanistic

step. This is followed by the oxidative addition of hydrogen.122 The

latter step is thought to be rate determining in the reaction but it

remains uncertain.65 Reductive elimination of aldehyde results in

reformation of the catalytic species.

Hydroformylation using modified rhodium catalysts results produces

a rather complex system. Many Rhodium species have been observed in

these systems including RhH(CO)(PPh3)3, RhH(CO)(PPh3)2, RhH(CO)2(PPh3)2,

and RhH(CO)2(PPh3).118 These complexes exist in equilibrium with each

other and are formed from either triphenylphosphine dissociation or CO

addition. The truly active species has not been determined but may be

any one or combination of these in solution.

The high cost of rhodium has limited its use commercially.

Development of a process which alleviates catalyst recovery problems

associated with conventional homogeneous systems while achieving high

selectivity and activity will result in an economically more viable

reaction process. A considerable effort has been devoted to modifying

homogeneous based rhodium catalysts to form heterogeneous catalysts.

This would offer immediate advantages in catalyst/product separations.

Use of insoluble supported catalysts in liquid phase reactions or vapor

phase reactions over supported catalysts might achieve this desired

objective.

Attempts to form heterogeneous oxo catalysts by attaching them to

supports through their phosphine ligands has not resulted in better

catalytic systems.123-125 Use of these supported catalysts in liquid







86

phase reactions leads to active and selective systems, however, catalyst

leaching occurs. Cleavage of the phosphorous carbon bonds used to

attach the catalysts results in degradation of supported complexes

providing a pathway for catalyst leaching from the support.126'127

An effective gas phase system is reported by Arai and

co-workers.128 The process utilizes silica gel as a support. It is

covered with a polymer formed from styrene and divinylbenzene and then

functionalized with phosphine ligands. Chlorocarbonylrhodium(I) dimer,

[Rh(CO)2C1]2, is then supported on its surface. This catalyst is found

to be effective in hydroformylating both ethylene and propylene at 100C

with atmospheric pressures of gaseous reactants. One of the major

drawbacks associated with this system is in the low selectivity obtained

for the production of n-butanal (63%) from propylene. The phosphine

linkage was viewed to be important since direct deposition of

(PPh3)2Rh(CO)C1 and RhC13 on silica exhibited little activity under

similar conditions. Deposition of (PPh3)2Rh(CO)C1 on alumina and

activated charcoal, however, produces active systems under substantially

more severe reaction conditions.129 Again low selectivities for the

linear aldehyde are obtained (56.5 to 65.5%). Gas phase

hydroformylation of ethylene and propylene have also been achieved over

zeolite supported rhodium catalysts.130-32 Again the selectivity of

these systems is poor (<67% linear aldehyde) and the simultaneous

hydrogenation of propylene and aldehyde is a severe problem.

The objective of the work presented in this chapter is to extend

what has been accomplished with membrane based reactors. This is

achieved by the development of a heterogeneous catalyst based system









which produces substantially increased activity while maintaining the

intrinsic benefits associated with the polymer membranes, namely their

selectivity. This is accomplished by deposition of a rhodium catalyst

and excess phosphine ligand on silica gel and its use in a gas phase,

continuous flow reaction. The development of a reactor which is

connected to a gas chromatograph ("on-line GC") results in an accurate

evaluation of the reaction properties. Modification of the catalyst

system by coating with a polar, nonvolatile system results in increased

activity and selectivity.



Experimental


The transition metal catalysts studied in this work are prepared

as previously described in chapter III. Identification and

quantification of reaction samples are conducted using a Perkin Elmer

model 900 gas chromatograph placed in conjunction with a continuous flow

gas reactor. This is more simply described as an "On-line" gas

chromatograph.

Catalyst Preparation

The catalysts employed in this study are prepared by dispersion on

a solid support in the presence of an excess quantity of

triphenylphosphine ligand. A typical description of this procedure is

detailed below. A 250 mL round bottom flask is charged with 0.600g of

silica gel (vacuum dried at 800C), 0.0919g (1.000 x 10-4 moles)

RhH(CO)(PPh3)3, and .262g (1.00 x 10-3 moles) triphenylphosphine.

Twenty five milliliters of chloroform is added to the solid mixture and







88

the solution stirred for approximately five minutes. It is then rotary

evaporated giving a yellow granular product which is vacuum dried

overnight at 450C. Catalysts which are coated with propylene carbonate

are prepared similarly with the propylene carbonate being dissolved in

the chloroform before its addition to the solid catalyst mixture. A

brief description of the preparation of all catalyst systems

investigated is listed below. Each is identified by itss appropriate

abbreviation throughout the remainder of the text.

RhH\SG

The catalyst is prepared by dispersion of 0.230g (2.50 x 10-4

moles) RhH(CO)(PPh3)3 and 0.656g (2.50 x 10-3 moles) triphenylphosphine,

PPh3, on 1.50g of silica gel using the method outlined above.

RhH\SG\PC

The catalyst is prepared by dispersion of 0.0919g (1.00 x 10-4

moles) RhH(CO)(PPh3)3 and 0.262g (1.00 x 10-3 moles) triphenylphosphine

and 0.08g of propylene carbonate on 0.600g of silica gel using the

method described above.

RhH\MPPS

The catalyst is prepared by dispersion of 0.0919g (1.00 x 10-4

moles) RhH(CO)(PPh3)3 and 0.262g (1.00 x 10-3 moles) triphenylphosphine

on 0.600g of macroporous polystyrene beads using the method outlined

above.

RhH\MPPS\PC

The catalyst is prepared by dispersion of 0.0919g (1.00 x 10-4

moles) RhH(CO)(PPh3)3 and 0.262g (1.00 x 10-3 moles) triphenylphosphine









and 0.08g of propylene carbonate on 0.600g of silica gel using the

method described above.

RhH\Nafion

The catalyst is prepared by dispersion of 0.0919g (1.00 x 10-4

moles) RhH(CO)(PPh3)3 and 0.262g (1.00 x 10-3 moles) triphenylphosphine

on 0.600g of ground nation using the method outlined above.

Rh(tfa)\SG

The catalyst is prepared by dispersion of 0.253g (2.50 x 10-4

moles) Rh(tfa)(PPh3)3 and 0.656g (2.50 x 10-3 moles) triphenylphosphine

on 1.50g of silica gel using the method outlined above.

Rh(tfa)\SG\PC

The catalyst is prepared by coating 0.92g of the previously

prepared Rh(tfa)\SG with 0.08g of propylene carbonate. This is

accomplished by rotary evaporation as described above in the catalyst

preparation section. This provides the exact quantities of reagents

used in the preparation of the other catalysts.

Reaction Procedure

The reaction is conducted in a stainless steel tube connected in

line with a GC as noted above. A schematic diagram of this setup is

shown in figure 4-2. The tube is 1/8" chromatography tubing, 45cm in

length with approximately a 25cm catalyst bed height. It is packed with

0.500g of the catalyst. A preblended lecture bottle of H2, CO, and

propylene is used to deliver the gases to the reactor. The flow and

internal pressure of the gases are controlled by valves positioned

before and after the reactor tube. Reaction temperature is controlled

by an oil bath. The extent of reaction is determined using a sample