- - -..L‘.: . I . ‘. .
I , r"; ‘I- _ -I fi'

MIMI ..Ww wk

. 1‘ .' 'q, '1'“. .1...“ 3‘ ’k‘ W'c‘tf‘fig

u? fih‘l‘s ," “93!?

   

 
 

I I.~"s":|.
I .‘ ‘1'
V I'; (NI 5.1 .' ‘

V'.‘

MmMW'
43‘ '0'- "‘ i“ .19,

      
         

  

      
  

  

 

          
         
  
   
 
 
  

~ "g1? -.
1' I ‘ :A. . '
.. 5.x" 1 r- '
d‘ V‘ ".H} 7 ‘n
qu'li‘gfl'a, .. .
., {2| K ‘ ' " 1
‘V. bu ‘ A Ibc‘
%’Ww;‘ 1'31 I
‘H . . I. ..
3‘1 ‘ " 11- ham-:41
‘H" I]\ t:_ I" ' I.‘.:J.V. .
.wa1flw

   
   

     

  

. Shh“ ‘»—‘.J\‘ .c» I
M 4 ‘ I _ I- . g‘i‘k . v. _ %‘%1_;".. ‘ -o‘xy:o-.y
. V“ 'I I}. ‘g'II'A‘lI ‘3. C. . I‘nI‘C'I “WK-"‘3'.“ “3.34?an "m ' ‘
' - ._ - 4p 0.! . . _
‘ {4* HIV. “13:53; 1%.? wt my..." $332354"? I'M?-
’ . .y‘ 7", - Xgfl-f‘ntg >' ‘n \-‘ 2.}: V‘VV'VV“ _. 23:51) RU 351;.” .r

 

'.' '.' ""1 ‘ $311353? :9 -;_,’I{{ Inc: JAV‘ ' ‘. V V ‘ILLifV-‘i'fzfir-V --‘_ _
N; V' H! W ‘56:??? flV P} E ‘ \‘r ‘zl 0|qu 3 fl 1", w ‘ “ ‘2‘ 33h
I: I; .-. .Y «‘3... “TV.

. . H“ . J‘
-. “I7;:'Sm* 91-49:.»
..’. a P L .6: E}!
«I ..".
4;;M3

  
  
    
       
   
     
   
 
  
   
 
   

   

Nib.“

 
  
 

‘v }II:’.'
T’\' I
.V.‘ fl];-
4- I "1'.
“I ‘3‘ b A

; . _ . Vii-«Ky, as “:44er

    
 

‘.
1\\. “I. . I“
,V I - . V
Pfihuw. VV%W’ a.
‘1: ”E. Effigy WV... Li'jZi I.. . ... 9-4

   
    
  

"J

.9.W1. ‘
a#;

p
v "I ‘ 4
‘ fi

W%fl
”~33?”
s:

   

 

ilk-g , -"
Aw 211,341 1:" E51:

1'
F!" .

     

|_ £525.“;
' '11‘
at.‘ .-
14 ".
r

«at - -. Arm

  
  

 

 

v‘ E‘

. i_ ?.
Em
"ti .

N
H
I

‘1

 

 

 

This is to certify that the

dissertation entitled

HIGHLY DISPERSED METALS IN PILLARED CLAYS.
PROPERTIES AS FISCHER-TROPSCH CATALYSTS

presented by
Edward Gordon Rightor
has been accepted towards fulfillment

of the requirements for

Ph .D . degree in Chemi strz

rofessor

Date NOV. 7! 1986

“(III-nu- ‘1"!!— o.’ . ~ '- IA ' 1. y - . 0.12771

 

 

 

MSU

LIBRARIES
-:_-—.

 

 

RETURNING MATERIALS:
Place in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

 

 

 

HIGHLY DISPERSED METALS IN PILLARED CLAYS.
PROPERTIES AS FISCHER-TROPSCH CATALYSTS

BY
Edward Gordon Rightor

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

1986

ABSTRACT

HIGHLY DISPERSED METALS IN PILLARED CLAYS.
PROPERTIES AS FISCHER-TROPSCH CATALYSTS

BY
Edward Gordon Rightor

Recently developed methods (1,2)'were utilized for the
synthesis of highly dispersed Fischer-Tropsch active metals
in the microporous regions of pillared clays. The stabili-
ty, activity, and distribution of hydrocarbons formed on the
resulting catalytic sites were investigated.

The first method (1) involved intercalation of hydroxy-
iron polycations into the clay galleries, followed by con-
version of the polycations to Fischer-Tropsch active sites.
The iron pillars produced primarily C1-C7 paraffins and
olefins, although minor amounts of branched hydrocarbons and
Cl-C3 alcohols also were detected. Microanalysis in a
scanning electron microscope indicated significant bulk iron
inhomogeneities and iron enrichment at clay edges. ‘X-ray
diffraction revealed that even though the Fe-pillared clay
was ordered and stable when calcined (air, 350°C), it was
XRD amorphous after catalysis or after storage in air for a
few months. This was most likely due to pillar hydrolysis.

In the second method (2), ruthenium carbonyls were

selectively dispersed in the micropores of alumina pillared

Edward Gordon Rightor

montmorillonite (APM). These Ru sites were catalytically
active without prereduction and gave non Schulz-Flory
hydrocarbon product distributions initially. Conversely,
prereduced Ru/APM catalysts did not show significant Schulz-
Flory deviations, but exhibited greater conversions and
enhanced production of high molecular weight products.

High resolution electron microscope studies (3) of
nonprereduced Ru/APM after 30 hrs of reaction proved that
Ru aggregated into large crystallites located on the exter-
nal surfaces of the pillared clay. In contrast, prereduced
catalysts gave a narrow distribution of small Ru microcrys-
tallites that were mainly retained inside the pillared clay
particles. .A model was proposed that explained the location
of relatively large crystallites (6nm) within void spaces
created by layer packing disorders. z-contrast imaging
showed small microcrystallites in thicker regions of the
clay which gave support for this model.

These Ru/APM catalysts exhibited high selectivities for
internal olefins and branched hydrocarbons, an unusual re-
sult for Ru-based Fischer-Tropsch catalysts. The yields of
these atypical isomer fractions were studied by variation of
reaction variables and the addition of l-olefin probe mole-
cules. These experiments gave support to the proposal that
this high isomerization selectivity is due to the strong
intracrystalline acidity of alumina pillared clay.

Edward Gordon Rightor

REFERENCES

l. Tzou, M.S.,' Ph.D. Dissertation, Michigan State
University, 1983.

2. Gianellis, E.P., Ph.D. Dissertation, Michigan State
University, 1984.

3. Rightor, E.G.; Pinnavaia, T.J., to be published in
Ultramicroscopy, 1986.

 

To Elizabeth McKenzie Keller and J. Gordon McKenzie

ii

ACKNOWLEDGMENTS

Dr. T. J. Pinnavaia has provided generous support and
encouragement along with the appropriate amount of guidance
during my graduate studies, for which I am deeply grateful.
He was always able to paint a broad picture of my PhJL
project, which helped considerably in surmounting the short-
range frustrations characteristic of research. His depth of
knowledge, enthusiasm for science, and amiable personality
are qualities I admire and I look forward to future
association with him.

I would like to express my gratitude to Dr. D. G.
Nocera for his dedication and flexibility in editing this
dissertation despite his hectic schedule. I would also like
to acknowledge Dr. H. A. Eick for helpful discussions.
Additionally, the support personnel in the chemistry
department were excellent and I appreciate their efforts.

The patience, approachability and generosity of Mr.
Vivion Shull, Dr. Karen Klomparens and Dr. Stan Flegler was
crucial in learning technical aspects of electron
microscopy methods, for which I would like to express my
thanks. Dr. John Heckman and Ms. Genevieve Macomber
provided additional technical assistance. Arizona State

University awarded me a scholarship to their NSF-sponsored

iii

"Imaging and Microanalysis School," which was greatly
appreciated.

Throughout my graduate studies my mother has been a
continual source of encouragement, support, and love which
has made it possible for me to presevere in this
undertaking. Kathy Smith, my fiancee, devoted many hours to
the word-processing of this dissertation and aided in the
processing of the electron micrographs. Her support,
inspiration, and dedication helped me make it through the
final stages.

Finally, I would like to thank my friends, previous and
present group members, and two super lab partners, Dr. Ivy
Johnson and Todd Werpy, for their encouragement and

understanding.

iv

TABLE OF CONTENTS

Page

LIST OF TABLES vii
LIST OF FIGURES viii
CHAPTER I - INTRODUCTION.. . . .. . .. . .. . .. . l
A. Constrained Systems. .. . . .. . .. . .. . l

B. Clay Minerals. .. . . .. . . .. . . .. . . 9
C. Fischer-Tropsch.. . . .. . .. . . .. . .. 20
D. Research Objectives. . . . . . . . . . . . . . 34
CHAPTER II - EXPERIMENTAL METHODS .. . .. . .. . .. 37
A. Clay Preparation . .. . . .. . . .. . .. . 37

B. Preparation of Alumina Pillared
Montmorillonite (APM).. . . .. . .. . .. . 38

C. Immobilization of Ruthenium Carbonyl on APM... 39
D. Catalysis. .... . . .. . .. . . .. . . .. 40
E. Electron Microscopy. .. . . .. . .. . .. . 48
F. Physical Measurements. .. . .. . .. . .. . 50
G. Chemical Analysis. .. . . .. . . .. . .. . 51
CHAPTER III - IRON-PILLARED CLAYS . . . . . . . . . . . 53
A. Introduction . .. . . .. . . .. . . .. .
B. Synthesis and Physical Properties. .. . .. . 62.
C. Fischer-Tropsch Catalysis.. . .. . .. . .. 67
D. Scanning Electron Microscopy .. . .. . .. . 78

E. X-Ray Diffraction. .. . . .. . . .. . .. . 92

F.

CHAPTER

G.

H.

Conclusions. . . .. . . .. . . . .. . . ..

IV - RUTHENIUM ON ALUMINA PILLARED
MONTMORILLONITE.. . .. . . .. . .. .

Physical Properties of APM and Ru/APM . . . .
Pristine vs. Prereduced Ru/APM Catalysts. . .
Electron Microscopy .. . .. . .. . .. . .
Nature of Ru Microcrystallites.. .. .. ..

Atypical Features of Ru/APM
Fischer-Tropsch Catalysts . . . . . . . . . .

Isomerization Mechanisms.. . .. .. . .. .

Alkene Probe Additions and
Hz/CO variation 0 o o o o o o o o o o o o 0

Conclusions . . . . . . . . . . . . . . . . .

APPENDIX A: Minimization of Microscope Artifacts. ..

APPENDIX 8: Feed Gas Impurities . .. . .. . .. . ..
APPENDIX C: Preparation.of Oxygen Scrubber.. . .. ..

REFERENCES

vi

94

98

104

106

114

133

139

154

160

175

179

183

188

190

Table

LIST OF TABLES

Shape Selectivity. Comparison of

hydrogenation on Pt/A1203 vs. Pt/ZSM-S,

from ref. 10. O O O O O O O O 0

Important Synthesis Gas Reactions.

from 75,70. 0 I O O O O O O O O

Adapted

Solubility Products of Fe(III) Oxides.

Adapted from 111, 25°C, Zero Ionic Strength

Sorption of Probe Molecules on Fe/PILC .

Summary of Physical Data for Fe/PILC

Summary of an Typical Fe/PILC FT
Product Distribution . . . . . .

Summary of Physical Data for
Al-Pillared Clays . . . . . . .

Ru/APM Catalytic Data at 1 atm .

vii

Page

22

58
58

58

66

103

110

Fi

10

11

re

LIST OF FIGURES

Channel structure of ZSM-5 and ZSM-ll
zeolites. Adapted from reference 6 ........

Molecular sieving of common zeolites.
Adapted from reference 6 ...................

The framework structure of montmorillonite,
an typical smectite. Adapted from ref. 16 ..

Schematic representation of pillared clay,
along with various classes of pillars. From

ref. 17 00......OI.OOOIOOCOOOOOOOOOOOCO0....

Influence of Process Variables on Fischer-
TropSCh OOOOOOOOOOOIOOOOOOOOOOOOCOOOOOOOIIOO

Fischer-Tropsch Reaction Mechanisms ........

Comparison (a) between a experimental non-SF
distribution (open symbols, 91), with a ESF
curve (closed symbols, 95). For curve (A)
Co/A1203 the mean pore radius was 6.5nm, and
in (B) it was 30nm. The particle size dist-
ributions used to generate the ESF plots are
given in (b). Reproduced from ref. 95 .....

Reactor setup for catalyst studies with
Fe/PILC 0..OOOOOOOIOOOOOOOOOOOOO0.0...O...00

Reactor system for catalytic character
ization Of Ru/APM 0.00.00.00.00...0.0.0.0...

Iron(III) hydrolysis and polymerization, as
shown by Flynn (111). The numbers in paren—
thesis refer to relative reaction rates.....

Change in the pH of Fe/PILC sols during the
process of washing (A), and XRD patterns (B)
of air dried FE/PILC films after selected
washings. The arrow in (A) indicates the pH
and washing at which flocculation first
occurred ...................................

viii

Page

11

11

22

25

32

41

45

57

64

Figure
12

13

14

15

16

17

18

19

20

21

22

23

Schulz—Flory plots of Fe/PILC hydrocarbon

production after 60 min. 0 ,103 min. A , andE

1259 min. of reaction. Reaction conditions

were 120 psi, 2750c, H2/CO=2, 2100 hr-l GHSV,

and 1.78 contact time .....................

Experimental hydrocarbon production.C>for
Fe/PILC (60min. TOS) compared to an cal-
culated (Schulz-Flory) distribution. A chain
growth probability of 0.42 was used for the
SF plot. Experimental data from Figure 12.

Experimental hydrocarbon production. C)for
Fe/PILC (1259min. TOS) compared to calcu-
lated [](Schulz-Flory) distribution. A chain
growth probability of 0.49 was used for the
SF plot. Experimental data from Figure 12.

Olefin/paraffin production for Fe/PILC cat-
alysts as a function of conversion for
C3i/C3- and C2-/C2- 00000000000000.0000...

Edge-on SEM view (a) of Wyoming Na+-mont-
morillonite air dried on glass. SEM image
(b) near a crack in the air dried clay......

SEM images of Fe/PILC air dried on glass at
various magnifications. The substantial

decrease in layer bending is evident at all
magnifications .............................

Generation of various signals in samples due
to interaction with the beam in electron
microscopes 000......OOOOOOOCOOOOOOOOOOOOOO.

SEM image of Fe/PILC (a) and the elemental
distribution (b) determined by an x-ray
microanalysis linescan .....................

SEM of a different area (a) and an accom-
panying linescan (b) between the arrows ....

SEM image of the undulating surface of Na+
montmorillonite (a) and an x-ray microanal-
ysis linescan (b) ..........................

Elemental microanalysis of Na+-montmor-
illonite between the arrows in Figure 16a .

Elemental linescan of a well ordered Cr-
pillared montmorillonite. Here the Cr dis-
tribution follows that of Si, indicating a
homogeneous pillared clay ..................

ix

Page

68

71

72

76

81

83

85

87

87

91

91

91

Figure

24

25

26

27

28

29

30

31

32

XRD patterns of Fe/PILC before and after
Fischer-Tropsch reaction. The initially well
ordered material (top) was essentially XRD
amorphous after 3 months in air.............

Arrangement of detectors in the specimen
chamber of a dedicated STEM. Reproduced from
ref. 148 0.0.0000......OOOOIOOOOOOOOOOOOOO

Infrared spectra of Ru/APM (a) immediately
after synthesis (2130, 2104, 2080 cm-l),
(b) prior to catalysis, aged 1 month in
dessicator, (2150, 2080 cm-l),(c) post
catalysis after 12.5 hrs, 2250C rxn. (2100,
2056 cm-l). a. ordinate expansion 400x ..

Schulz-Flory plots for Ru/APM catalysts after
8 min. C) , and 138 min. [3 of Fischer-
Tropsch. The catalyst was not prereduced
prior to reaction. Reaction conditions:
2750C, 8 atm., 300 hr-l, H2/CO=2 ........

Comparison of Schulz-Flory plots for
prereduced C) and nonprereduced [3 catalysts
at 275oC. A prereduced catalyst run at 2250C
gave the distribution shown by [j . Reaction
conditions: 1 atm., 300 hr-l, 50 min. TOS.

Dark field image of Ru/APM catalyst after 30
hrs. of catalysis at 2750C showing Ru
aggregates. The catalyst was not reduced
prior to catalysis. Arrows indicate agg-
regate examJned by the EDS analysis in
Figure 33 .................................

Dark (a) and bright field (b) images of the
Ru/APM catalyst showing a coalesced Ru
particle. Catalyst was not prereduced ....

Twinned Ru aggregate on a Ru/APM post-run
catalyst sample not reduced prior to
catalysis. The catalyst was run for 30 hrs.
of FT catalysis ............................

Ruthenium crystallite sizes observed by STEM
for nonprereduced catalysts after 30 hrs. of
FT reaction at 2750C. Here ni is the number
of particles with diameter di, dn is the

number average and ds is the surface average
particle size ..............................

X

Page

93

103

105

108

113

117

117

117

118

Figure

33

34

35

35c

36

37

38

39

40

EDS spectrum from the Ru crystallite indi-
cated in Figure 29. The Cu peaks are arti-
facts originating from the Cu microscope

grids OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

Ru crystallite sizes determined by STEM
studies on numerous Ru/APM particles after 30
hrs. Of reaCtion OOOOOOOOOOOOOOOOOOOOOOO

Representative prereduced Ru/APM catalyst
particle, found by STEM bright field (a), and
dark field images (b), showing small Ru
microcrystallites. An EDS microanalysis is
given in Figure 35c. The catalyst was run
for 30 hrs. of FT at 2750C and 1 atm .....

EDS spectrum from the 5 nm microcrystallite
indicated in Figure 35 OOICOOOOIOOOOOOOOOOO

Model of air dried pillared clay showing
mesopores created by layer packing disorders
and layer bending. Plausible locations of Ru
microcrystallites >2nm (labeled Ru), and less
than 2nm (filled triangles) are indicated.
Slabs represent clay layers and small open
circles the "alumina" pillars .............

Pore size distribution of APM from liquid
nitrogen desorption isotherms. Here rp is
the mean pore radius, and Vp is the change
in pore volume .............................

Dark field (a) image of Ru/APM after catal-
ysis. The area indicated in (a) is shown
below in (b), using the z-contrast ratio
imaging technique, which shows microcrystal-
lites < 5nm. Catalyst prereduced ..........

Bright field (a) and difference z-contrast
image (b) for a prereduced Ru/APM catalyst
particle showing the ability of z-contrast to
image Ru in thicker clay regions ........

STEM bright field image (a) for prereduced
APM following FT catalysis. Nanodiffraction
patterns (b) and (c) are from the micro-
crystallites indicated. (Provided courtesy
of J. M. Cowley, Arizona State University).

xi

Page

 

120

120

123

124

130

131

135

135

138

Figure
41

42

43

44

45

46

47

Representative gas chromatograms for
prereduced Ru/APM catalysts showing
production of branched and internal olefin
products. Chromatogram (a) is for the
gaseous fraction, and (b) illustrates the
components in the condensed liquid ........

Variation of the branched/straight chain
hydrocarbon selectivities with temperature
for Ru/APM catalysts. The symbols represent
C4 0 , C5 D , C6 A and C7 6) hydrocarbon
fractions. The FT conditions were; 120 psi.,
1910 hr-l, H2/CO= 2, 50 min. TOS, prereduced.

Change in the branched/straight hydrocarbon
ratio ,normalized for conversion variation,
at various pressures. Here the [3 rep-
resent data at 1 atm, Aat 70 psi , and Q
at 190 psi. Intermediate pressures follow
these trends but are excluded for clarity.
Reaction conditions; 2500C, 1910 hr-l, H2/CO
32,50 min TOS ..............................

Production of Branched/straight hydrocarbons
at longer run times (100 min.) for C4 0. El C5
C6 [3 and C7 C) hydrocarbons. Run
conditions were the same as in Figure 42 ...

Steady state (200 min.) selectivity for
branched hydrocarbons for C4 Q , C5 E] , C6 A
and C7 C) products. Run conditions were
the same as in Figure 42 ...................

Effect of space velocity on the branched/
straight ratio. The C4 hydrocarbons are
represented byG) ,C5 by E] , C6 by A , and
C7 are shown by C) . The slashed symbols
show the ratio when the flow was returned to
the original value. (2750C, 110 psi,
H2/C0-2, 45 min TOS, prereduced, contact time
from 1.3-0.33 seconds) ................

Production of internal olefins relative to
terminal olefins (i Cns/T Cn-),and n-alkanes
(i Cns/ Cn-) as a function of temperature.

Here the symbols are defined as: iC4=/Tc- is(3

,ic4s/c4- <5) , ics-rrcs- A , and 1cs-/cs-
is E]. Conditions were 120 psi. , 1910 hr-l,
H2/CO=2, 50 min. TOS, prereduced ........

xii

Page

140

142

144

146

147

149

151

Figure

48

49

50

51

52

53

Page

 

Comparison of total alkene/alkane , norm-

alized with respect to conversion, for 1 atm.

070 psi A , and 190 psi E] . FT

conditions were: 2500C, 1910 hr-l, H2/CO=2,

50 min. TOS, prereduced .................... 153

Ratio of alkenes to alkanes for Ru/APM cat-

alysts as a function of flow rate. The

symbols show C4 C9 , C5 E] and C6 A
hydrocarbons. The slashed symbols show the
recycle values. Conditions were 2750C, 110

psi., H2/CO=2, 45 min. TOS, contact time from
1.3-0.33 seconds ....................... 155

Yields of internal olefins relative to 1-

alkenes and n-alkanes as a function of flow

rate. Symbols are; iC4=/1-C4= C)

1c4-/Cn- (a ,1 c5-/1-c5=A and iC5=/Cn- E]

(2750C, 110 psi, HZ/CO=2, 45 min. TOS, 1.3-

0.33 sec. contact time) ..................... 156

Plausible mechanisms to account for hydro-

carbon branching showing: (a) insertion of

propene, (b)addition of chemisorbed methyl

species, (c) isomerization associated with

hydride shift, (d) formation and rearrange-

ment of an alkylidiene, (e) carbocation
rearrangements .............................. 159

Increase in hydrocarbon production resulting

from incorporation of 1-alkene probe

molecules. The symbols signify: steady state

FT (100 min.)(3 ,hydrocarbon distribution

during the first alkene addition A, and

after the second addition [3 . The [alkene]

in the second addition was twice the first.
(2750C,110psi.,4336 hr-1,H2/CO=2, prereduced. 162

Production of branched and straight hydro-

carbons before and after l-alkene additions.

The A show steady state FT (180 min.) , O

and [3 production after the first and second
alkene additions, respectively. The slashed
points are for straight isomers and the

regular symbols are for branched products.

(2750C, 110 psi., 4336 hr-l, H2/CO=2,

prereduced) ............................... 163

xiii

Figure Page

 

54 Breakdown of the l-olefin probe molecules
added during FT into internal olefins (iCn=),
and total paraffins (Cn-). The regular
symbols for FT steady state (180 min.) C) ,
first addition A ,and second E] refer to
paraffins and the slashed symbols to internal
olefins. FT conditions as in Figure 53..... 165

55 Transformation of 1-alkene probe molecules to
alkanes in the absence of CO (no FT) as a
function of the alkene to hydrogen ratio.
The symbols represent: C) C2, [3 C3, /\.C4,C)
cs, El C6, and c7. (2750c, 1 atm,
3230 hr-l) ................................ 168

56 Transformation of the added l-alkene probes
to internal olefins as a function of the
alkene to hydrogen ratio. The symbols are;
cisC4 G , trans C4 9’, cis C5 A, A trans C5
1-C4-(),1-C5=/"\. (2750C, 1 atm., 3230 hr-l,
prereduced) ........................ 169

57 Gas chromatograms of l-alkene probe molecules
after passing through the Ru/APM catalyst
bed. (a) 250C before any isomerizations, (b)
reactor effluent when alkenes are added to
the catalyst at 2750C , 1 atm., 3230 hr-l, 50
min. TOS, (1- Cn-/H2=1.25) , (c) 1-
Cn=/H2=2.8................................ 170

58 Change in the selectivity for branched
hydrocarbons during FT with H2/CO. The
symbols are: 0 04, A cs, El C6, 0 c7,,'\ ca.
(2750C, 120 psi.,3230 hr-l, prereduced). 172

59 Isomerized hydrocarbon production vs. normal
FT production (n-alkanes and l-alkenes) as a
function of the H2/CO ratio. Here C4
hydrocarbons are shown as (D , C5 as ,
and C6 as (A . (2750C, 120 psi., 3230hr-1). 173

60 STEM dark field image of Ru crystallites
formed by beam induced agglomeration in the
microscope. Microdiffraction patterns are
from the pseudo-hexagonal (b), and rect-
angular crystallites. Sample stage at room
temperature ................................ 181

61 Ru/APM after 30 hrs. of FT reaction with feed
gas containing carbonyl impurities at low
(a), and high magnification (b). Microscope
sample stage at room temperature............. 185

xiv

Figure Page

 

62 EDS analyses of the aggregates indicated in
Figure 61, which show the presence of Fe and
Ni impurities 00.0.0000.........OOOOOOOOOOOO 186

XV

Chapter I

Introduction

A. Constrained Systems

Chemical research involving the synthesis, character-
ization, and applications of constrained reaction systems
has grown dramatically since 1960 into a broad, multidis-
ciplinary area. The principle objective is to geometrically
alter the local environment of molecules to induce unusual
physical or catalytic properties. This can be accomplished
by confining a guest molecule within a somewhat rigid host
matrix; The reactivity of the guest is thus limited and
interaction with reactant molecules may afford greater se-
lectivity than obtained with free molecules. Suitable or-
ganic constrained systems can involve molecules or polymers
that mimic enzymes, whereas inorganic systems are commonly
materials with two or three dimensional pore networks. The
diversity of materials being used to investigate the effects
of molecular constrainment can.be appreciated by considering
the numerous classes of constrained systems which includes:
cyclodextrins (1, 2): lipids and microemulsions (2): bi-
layered systems and graphite intercalants (1, 2, 3):
zeolites and layered clays (2, 3); chalcogenides and

,B -aluminas (2, 3): and crystalline silicic acids (2, 3).

1

Some of the most effective inorganic constrained sys-
tems have been zeolites. Zeolites have demonstrated remark-
ableselectivity in transformations of organic molecules and
provide an excellent example of the types of geometrical
influence that many inorganic constrained systems strive to
achieve. The synthesis and technical application of
zeolites started with the first industrial research efforts
in 1948 at Union Carbide, and has generated a new field of
inorganic materials with vast scientific interest. In the
twenty-five years following the commercial application of
zeolites in 1954, over 15,000 scientific contributions and
over 10,000 patents have explored this area(4). In 1979
the molecular sieve industry alone, was projected to be a
250 million dollar market (4).

Among the attractive characteristics of zeolites for
industrial catalysis are their sharply defined pore sizes,
high and adjustable intracrystalline acidity, and high sur-
face areas. They also have excellent hydrothermal stabili-
ties and their relatively easy synthesis from inexpensive
raw materials makes them economically attractive.

Zeolites are crystalline aluminosilicates composed of a
three-dimensional network of linked silica and alumina tet-
rahedra. The tetrahedra link through shared oxygens to form
rings typically containing 4-12 tetrahedral units. These
ring structures form the entrances to channels that define
the diameter of pores that propagate throughout the

structure (5). The size, and arrangement of the channels

resulting from connected pores are of utmost importance
as they affect many unusual properties of these materials.
Channel networks for two pentasil zeolites of particular
catalytic importance (ZSM-S and ZSM-ll), are shown in Figure
1. The dimensions of zeolite pores and channels are similar
to the critical dimensions of simple isoparaffins and sub-
stituted aromatics. 'The relationships between zeolite pore
dimensions and hydrocarbon reactant, intermediate or product
size then provides the fundamental basis for shape selective
catalysis. The size and shape of the pores and channels
influence the sorption of molecules into zeolite interiors.
Figure 2 shows a comparison between zeolite window (pore)
size and the kinetic diameters of representative sorbate
molecules “fl. Molecules with critical dimensions less than
those of the pores can pass through these openings whereas
molecules that are too large cannot. This leads to selec-
tive sorption of molecules, termed molecular sieving (7).
The shape selectivity of zeolites in catalysis, first
reported by Weisz and Frilette H”, has become a dominant
feature in catalytic applications. 8. Csicsery (9), in a
recent review on zeolite shape selectivity, distingushed

four different types of selectivity :

(a) Reactant selectivity occurs when the pores of windows
are too small to allow some reactant molecules to diffuse to

active sites.

The hydrogenation of olefins and aromatics on Pt/ZSM-S

 

 

 

 

 

 

 

 

 

ELLIPTICAL CHANNELS CIRCULAR CHANNELS

(5.1 x 5.6 X) (5.4 x 5.6 A)

Figure 1. Channel structure of ISM-5 and ZSM-ll zeolites.
Adapted from reference 6.

 

9- (CcFe)3N

 

8..—(C4|Ee)3N

 

6 :— Neopentane
”‘Benzene

5--Isobutane

 

:‘Propane
_.CO/Cflc
:QNz

3-1 02

-\E20

 

 

 

 

 

 

Pore or molecular diameter (A)

CaX
Mordcnite

2‘

Mordenite (3.311 pores)
A

NaY

NuX
Offrctitc
ZSM 5
7.00] itc -1-
Erionitc
Chabasite
CaA

NuA

LJA

KA

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Zeolite type

Figure 2. Molecular sieving of con-on zeolites.
Adapted from reference 6.

5
provides a dramatic example of reactant selectivity.
Table 1 shows that bulky hydrocarbons which were reactive on
macropore Pt/ A1203, were almost inert on the zeolite
catalyst as they were excluded from the active sites.

Table 1. Shape Selectivity. Comparison of Hydrogenation
on Pt/Alea vs. Pt/ZSM-S, from ref. 10.

 

Catalyst Temp.(°C) Pt/Alea Pt/ZSM-S

Hydrogenated (x)

 

Eexene 275 27 90
4,4-Dimethylhex-l-ene 275 35 (1
Styrene 400 57 50
Z-Hethylstyrene 400 58 (2

 

This reactant selectivity has been illustrated by Csicsery

as follows.

’ ///
W .../..__. M/

7///////////////.
W DV/l/l/l/ll/lll

Diffusion is of paramount importance in shape selective
catalysis, as one type of molecule will generally react
preferentially if its diffusivity is one or two orders of
magnitude greater than competing types of molecules (9).
Branching has a profound effect on diffusivity, whereas
increased hydrocarbon length results in a smaller diffu-

sivity effect .

(b) Product selectivity results when the diffusion of cer-

tain products from active sites to the external surface of

the zeolites is limited.

Diffusion limited molecules may produce an equilibrium shift
to less bulky molecules or can block the pores resulting in
catalyst deactivation. The diffusivity of paraxylene is
approximately 103 greater than m-xylene in ZSM-S (10).

This product selectivity is illustrated below.

CH,0H.©.___. ($3.6; ) __._©_

 

 

The selectivity can be further modified by the addition of

phosphorus which may partially occlude the pores.

(c) Restricted transition state selectivity occurs when
steric constraints of the rigid zeolite pore limit the
formation of bulky intermediates or transition states.

The tendency for selective zeolites with small pore openings
to exhibit greatly decreased coking rates probably is the
result of bulky, fused ring polycyclic hydrocarbon formation

being sterically prohibited in the pore confines.

(d) Molecular traffic control can occur when more than one
type of pore system exists in zeolites.

Adsorption data from Derouane (11) presents the possibility
that reactant molecules could enter the cylindrical, zig-zag
pores of ZSM-5 whereas diffusion of bulkier molecules would

preferentially occur through the linear elliptical pores

(see Figure 1).

These various modes of shape selectivity make zeolites
versatile and effective supports for catalytic hydrocarbon
transformations.

The intracrystalline acidity of zeolite channels is
another prominent feature responsible for the unusual capa-
bilities of these materials. Acid sites are generated in
zeolites by exchanging the natural cations with more polar-
izing ones (higher charge/radius) or with ammonium ion
followed by calcination. The greatest Bronsted acidity is

attained between 400 and 550°C,

4' H O
0 § . 2
I, II‘ E
O
0\51,0\- ’0 [uu‘ is“: o\ 31/ 0;» ,0 coo-soo- o\u / 31/0
—
0/‘00’\0 o’s’n‘oo<0 0/‘oo"

 

 

but above 550°C water is expelled and primarily Lewis acid

sites are formed.

a+
( °s\ ”/0\ .°/ /°) ’550": (°\s/o\’;(°\s 51(0 0;”; oO'H\:)

1 /
0”5 “0 0’ ‘0 0’; “0 0 ‘° ° ° °

 

Bronsted acidity is responsible for the carbocation chemis-
try of catalytic cracking and many other hydrocarbon
transformations.

Acidified zeolites have much greater activity in these
reactions than amorphous alumina-silica mixtures or acid
treated clays. This is due to zeolitic acid sites being in
greater concentration and proximity within the pores, and
more uniform in their intermediate acid strength (12,14).

Further, the concentration of hydrocarbons within the pores

8

is typically much higher than in silica-alumina pores. .Add-
itional information on surface acidity is given in the excel-
lent review by Corma and Wojciechowski (12) and the book by
Gates, Katzer and Schuit (13).

Yet zeolites do have limitations. .Although the number
of different zeolites synthesized is rather large, relative-
ly few zeolites (X, Y, ZSM-S, Mordenite, Erionite, A) have
found widespread applications. This is partially due to the
fact that the small pores of zeolites severely limit their
application in reactions that involve reactants with molecu-
lar dimensions larger than the pore openings. The super-
cages of faujasitic zeolites (X and Y) represent the largest
zeolite pores at about 7.5.A. For example, heavy vacuum gas
oil, which contains C18-C25 aromatics, has limited accessi-
bility to zeolite pores and cannot be effectively cracked to
more useful hydrocarbons. Typically large molecules are
cracked on zeolite exteriors which results in a substantial
loss in selectivity (13).

Partially in response to this zeolite limitation there
is considerable interest in microporous materials that fea-
ture more adjustable, larger pores, but also have high
intracrystalline acidity similar to that of zeolites. Ad-
justable pores would allow more flexibility in influencing
the catalytic chemistry of larger molecules. Materials with
great potential in this regard, that also offer molecular
constrainment and intracrystalline acidity, are layered

clays.

B. Clay Minerals
1. Structure

The term clay mineral defines a large, diverse class
of inorganic materials that have particles less than 2
microns in diameter. Elemental units of this member of the
phyllosilicate family are silicon-oxygen tetrahedra, and
aluminum-oxygen octahedra. The tetrahedra and octahedra
link in various fashions to form structurally distinct
members of each clay mineral class.

The mineral class of greatest interest here is
smectite clays which are 2:1 layered structures containing
one octahedral sheet sandwiched between two tetrahedral
sheets. .As formed.in nature a variety of cations may be
found in place of the aluminum and silicon in octahedral
and tetrahedral layers of idealized clays. This results in
numerous clay isomorphs. For example in montmorillonite
Mg2+ fills some octahedral cation positions, and in bei-
dellite Al3+ is found in tetrahedral positions. In both
cases a charge deficiency results as the Mg2+ and Al3+ have
lower charges than the cations normally occupying these
positions. This charge deficiency results in a net neg-
ative charge on the layers that is balanced by hydrated
cations between the layers. The magnitude of the layer
charge significantly affects chemical and physical proper-
ties of these clays and can be used in their classification
(15) .

Smectites have a layer charge between the low

10

charged pyrophillite-talc group and the highly charged
vermiculites and micas. The smectite layer charge can
arise in the tetrahedral layer such as in beidellite and
saponite, or in the octahedral layer as in hectorite and
‘montmorillonite. Figure 3 illustrates the smectite frame-
work, origin of layer charge, position of interlayer hy-
drated cations, and d001 spacing measured by x-ray diff-
raction. The d001 spacing is the sum of the layer thickness
and free space between layers which is also called the

gallery height.

2. Smectite Properties

The intermediate layer charge of smectites is respon-
sible for many of their unique physical properties.
Highly charged clays (fl; micas) tightly hold interlayer
cations, while low charged clays (§§L_talc) require few
interlayer cations, but smectite interlayer cations can be
readily exchanged by a great variety of cations. Addi-
tionally, smectite layers can.be separated by multilayers
of polar solvents so the layers can be swollen to accom-
odate large molecules. Their considerable cation exchange
capabilities and swellability give these minerals their un-
paralled versatility for intercalation of almost any
desired cation (17). Intercalation here means insertion of
cation guests between the layers of host (clay). Inter-
calation of vermiculite, which has a relatively high layer
charge, commonly yields "stuffing" of the interlayer with
cations which leads to diffusion related problems in

 

 

 

Figure 3. The framework structure of montmorillonite, a
typical smectite. Adapted from ref. 16.

 

{M- (P? 7P;

L (5+) ea: ‘

 

 

 

]
n “I
I
+ nt
R/ XV: (T? E M" (0H),
R ,1,+ “Merle”;+

Figure 4. Schematic representation of pillared clay,
along with various classes of pillars. From ref. 17.

12

catalysis.

3. Catalysis using Expanded Clays

The ability to swell smectites to accomodate different
sized guests, intermediate cation exchange capacity, and
large interlamellar surface area (750 mz/g, theoretical),
suggests that these materials should be ideal for heteroge-
nizing homogeneous catalysts. The goal of heterogenizing a
homogeneous catalyst is to combine the advantages of het-
erogeneous and homogeneous catalysts while minimizing their
disadvantages (18-20).

Intercalation of catalytically active reactants be-
tween silicate sheets then may result in product selectiv-
ities different from those encountered in homogeneous cat-
alysis. For example, Pinnavaia and coworkers (21) dem-
onstrated that Rh(PPh3)2+ intercalated hectorite gave
significantly improved yields of cis-alkenes in alkyne
hydrogenations relative to isomerization products. This
work also showed that choice of swelling solvent can dic-
tate gallery height and influence reaction rates and sel-
ectivities relative to homogeneous catalysts. The size and
shape of alkyne substrates also proved to be important,
leading to the postulate that transition state size select-
ivity was present.

Hydrated interlayer cations are more acidic in clays
than commonly found in solution (22,23) which is another

important property of intercalated clays. Although

13

intercalated clays can exhibit both Lewis and Bronsted
acidity, protonic acidity is dominant in reactions with
organic substrates. Bronsted acidity originates from the
polarization of water molecules coordinated to cations
restricted between the clay layers, and increases with
increasing cation charge to radius ratio (24,25). Further,
as the interlayer water content and layer separation
decrease, the cations are more confined and hence Bronsted
acidity increases. The relative state of hydration and
nature of interlayer cation present then greatly influences
the acidity, which is important to the unusual chemical
reactivities, conversions and selectivities.

Natural montmorillonites, have Hammett Ho values from
1.5 to —3 (26). Dramatic increases in acidity can be
achieved by simply replacing the natural cations with pro-
tons, resulting in acidity values -5.6 and -8 (27). The
surface acidities of such modified clays are comparable to
that of concentrated nitric and sulfuric acids. The Bron-
sted acidity of these clays is the key to producing carbo-
cations which are responsible for the transformations,
rearrangements and conversions observed with organic mol-
ecules.

The versatile nature of clays in catalyzing unusual
conversions of organics has been the subject of many recent
reviews (28-30). Adams gfiiggfi (31) have reported the pseu-
docatalytic selective conversion of straight chain alk-l-

enes over ion-exchanged montmorillonites to

ll:

di(2,2'-alkyl) ethers. The interlayer cations that yield
the greatest acidity, Fe3+ , Cr3+ , and Al3+, also are the
most reactive, but montmorillonite intercalated.with Cu2+
gave the highest selectivity. This reaction gives unusual
products in that rearrangement products characteristic of
other ether syntheses are avoided. Dehydration of primary
alcohols in acidic homogeneous solutions normally yields
fast rearrangement of a primary carbocation to the more
stable secondary carbocation. The corresponding alkene is
the favored product. Yet, Ballantine gt ag: (32) found
that primary alcohols are converted to di(alk-l-yl) ethers
over exchanged montmorillonites.

For the 8N2 reaction mechanism to prevail on pro-
tonated primary alcohols without rearrangement the react-
ants must be in proximity. The authors suggest that the
clay gallery promotes reactant contact and the intermolec-
ular product is preferred (ether) over the intramolecular
product (alkene). Weiss (33) in fact proved that reactant
pair proximity can effect catalytic conversions in clay
galleries. By using a series of smectites with varying
layer charge, he was able to change the distance between
gallery cations which altered product selectivies in oleic
acid oligomerization. Thus the acidity of and distance
between interlayer cations are variable components of clay
catalysts which can afford adjustable environments for

unusual chemical conversions.

15

4. Pillared Clays

The collapse of clay layers resulting from interlayer
dehydration at elevated temperatures severely limits appli-
cation of these layered clays in heterogeneous catalysis.
The demonstration by Barrer and McLeod (34) in 1955 that
interlayer porosity could be stabilized by the intercal-
ation of molecular props or pillars however, has led to
materials that successfully circumvent this problem. These
materials, called pillared clays, contain robust molecular
props that hold clay layers apart at elevated temperatures
so that interlayer porosity is retained (Figure 4). The
adjustability of pillared clays, and the fact that their
pores can be larger than zeolites, has led to renewed
interest in clays as heterogeneous catalysts. This is
particularly true for reactions with large molecules, such
as high molecular weight crude oil.

Barrer and McLeod's early work involved the intercala-
tion of N’(CH3)4+ and N(C2H5)4+. These cations provided
stable intracrystalline porosity so that both nonpolar and
polar molecules could be adsorbed in intralamellar regions.
Selective sorption of pentanes with these clays was inver—
sely related to their cross-sectional area, i§5_uptake of
C5H12 > iso-C5H12 > neo-Csle. Also uptake of 02 increased
eight-fold and C6H6 adsorption increased twelve-fold, com-
pared to similarly prepared Na+ montmorillonite. The
sorption properties of a variety of methyl ammonium

'montmorillonites (ie. (CH3)4_XNHX)'+ were studied by Barrer

16

and Reay (35). The found exclusion of molecules with
critical dimensions greater than the gallery heights of
those expanded clays, and determined that alkyliammonium
pillars were stable up to approximately 250°C.

The concept of pillared clays was further advanced
with the intercalation of a large bicyclic amine called 1,4
diazabicyclo [2,2,2] octane (DABCO). Mortland and Berk-
heiser (36) showed the protons of diprotonated DABCO to be
labile and observed conversion of acetonitrile to acetamide
with DABCO-smectites. Shabtai 2E.£l;(37) demonstrated
that DABCO-montmorillonite had higher catalytic activities
for esterification of carboxylic acids than alkyl ammonium
intercalated clays. They also demonstrated that these
materials had shape selective capabilities.

The third major class of molecular props to be inter-
calated in smectites is the tris-metal chelates. .A variety
of metals and coordinating ligands have been used including
Cu2+ and Fe2+ complexed with 1,10-phenanthroline (38),
Fe2+, Cu2+, Ru2+ with 2,2-bipyridyl (38,39) and Cr3+,
Co3+,Cu2+, with tris-ethylenediamine (40). The ability of
these materials to bind more metal complex than expected,
called intersalation, follows the anion tendency to ion
pair (Li 804' > Br' > C1" ).

The report by Brindley and Sempels in 1977 (41) of
smectites pillared with hydroxy-aluminum cations led to a
new class of polynuclear metal pillared clays. The hydroxy

aluminum beidellites prepared had basal spacings near 17 A

17

and maintained this spacing upon calcination up to 500°C.
iMontmorillonites pillared with hydroxy cations of Zr (42)
showed similar basal spacings (d001), thermal stabilities
and surface areas (300-400 mz/g). Nitrogen adsorption iso-
therms for these materials showed considerable micropore
character and displayed Langmuir adsorption isotherms.
Lahav gt 211.643) reported the basal spacing of Al-pillared
clays to be dependent on age of the hydroxy-Al solutions,
OH/Al ratio, and concentration of A1 solution relative to
clay. Vaughan and Lussier (44) found that Al-pillared
clays adsorbed 1,3,5-triethylbenzene (7.6 A) and smaller
molecules, but not 1,2,3,5-tetramethylbenzene (8 A) or per-
fluorotributylamine (10.4 A). In addition to demonstrating
these molecular sieve properties, they also found that
mesopores (60-150 A diameter) were present along with the
dominant micropore (<15 A) character. Further, they sugg-
ested that gallery porosity could be optimized by control-
ling ion exchange capacity of the clay, degree of pillar
hydrolysis, and drying conditions.

The potential of these clays for heterogeneous cataly-
sis was demonstrated by Shabtai and Lahav (45,46) and
Vaughan e_t _a_l; (47,48). Pillared clays showed improved
reaction rates for catalytic cracking of hydrocarbons
(especially those with kinetic diameters > 9 A) (49), and
produced gasoline octane ratings in the cracking of Gas Oil
comparable to zeolite catalysts (47).

A variety of cations have been used since these

18

initial studies of polynuclear metal pillared clays. In
addition to further developments with Al (50-52), and Zr
(52-54) pillars, Si (55), Bi (56), Cr (57),Ni (42), and Fe
(58,59) pillars have been used to induce permanent clay
porosity. ‘Yet, the A1 and Zr pillars seem to be the most
thermally stable pillars found so far.

Two general synthetic routes are commonly followed for
the pillaring of clays: in gitg construction of the pillar
precursor within the galleries, and exchange into the gal-
leries of a pillar preformed in solution (60). The 12
gitu method is illustrated below with a silicon acetly-

acetonate (acac) pillar.

 

Si(acac)3+ + Na+ —-—) Si(acac)3+ + Na+

 

 

 

Si(acac)3+ + azo—-) snot!)4 + a“

+ 3(C83CO)2CH2

 

 

Although this method circumvents the numerous manipulations
common to the exchange method, it does not allow control of
the pillar stoichiometry. Typically pillar hydrolysis
continues until the interlayer is filled, producing a
chloritic-type layer. This method can be used however, to
form unusual pillars within the galleries such as Ta and Nb
pillars (61).

Direct exchange of pillars formed in solutions into
clay layers is the preferred synthetic method for hetero-
geneous catalyst preparation as it yields materials with

high porosities. The stoichiometry of the cluster and

19

density of the pillar within the gallery is readily
controlled with the direct method. .A variety of cation
pillars can be introduced this way, including a new pillar
based on silsequoxides (62,63).

5. Properties of Metal-Hydroxy Pillars

Salts of metallic elements with high charge to radius
ratio polarize water molecules and induce the loss of pro-
tons when dissolved. The addition of hydroxide ions will
facilitate this process, and the subsequent formation of
polymeric species. This process, termed olation, is depic-

ted below.

8).

(
[Mill 2..°’-1.°“'” 1’ ——. —E230iu< ;>um20)‘_]’é;_; [)u’a ‘0 ’"<

Elevated temperatures, longer aging times, and higher pH
promote the hardening of the polymers formed through

another process called oxolation.

[5»?

(x-4l*

>H€:[ --‘. [M’D‘~nl’o‘~ + 43‘

[In \OI \qu \

3-0( )-0

For high surface area pillared clays it is desirable
to intercalate small, yet robust, oligomers. These can be
obtained if the above mentioned variables are carefully
controlled. Once inside the confines of the clay galleries
however, additional hydrolysis can occur, again due to the
polarizing nature of the layer. Thus, once clay solutions
are pillared the clay is quickly washed to remove excess

electrolyte and dried. Following drying the pillared clays

20

are normally calcined to convert the hydroxylated pillar to
the more stable oxide as shown below for alumina-pillared

clays.

 

 

- (A
All304(OH)24(H20)1('2’-§)+——’ ("A1203") + (7»le+

 

 

Pillar dehydroxylation also results in the formation of
Bronsted acid sites. These sites are in large part respon-
sible for the catalytic transformations observed with pil-
lared clays from polynuclear metal cations below 400°C.
Above 400°C, the Lewis acidity of the pillars begin to
play a more dominant role.

Although the acidity of pillared clays has been cen-
tral in the previously described transformations of organic
molecules, these materials can also be loaded with catal-
ytically interesting metals to take advantage of the micro-
porous nature of this unusual support. This can be dem-
onstrated in Fischer-Tropsch chemistry, as described in the

following pages.

C. Fischer-Tropsch (FT)

Progress in the catalytic synthesis of hydrocarbons
from C0 and H2 began in 1902 when Sabatier and Senderens
(64) found that reduced nickel catalyzed the hydrogenation
of C0 to methane. iLater, workers at BASF showed that alco-
hols, aldehydes, ketones, fatty acids and assorted ali-
phatic hydrocarbons could be formed over alkali promoted

cobalt and osmium oxides (65). Fischer and Tropsch (66,67)

21

went on to demonstrate the production of hydrocarbons > C5
at one atmosphere using Fe catalysts, and greatly advanced
the fundamental understanding of this reaction.
Commercialization of this process soon followed and by

1943 the ten German plants alone produced 585,000 metric
tons/yr. of hydrocarbons (46% gasoline, 23% diesel, 3%
lubricating oil, 28% waxes and detergents) (68). Scientific
and technogical interest in FT reached a high point between
1940-1955, as reviewed in a number of important papers (69-
71), before the discovery of inexpensive Middle East oil
reserves caused a shift towards an oil based fuel and
chemical industry. Industrial research and application of
FT has been cyclic ever since, and the great upturn in oil
prices in 1973 resulted in renewed research and development
efforts in this area.

The wide variety of organic products formed in the
catalytic hydrogenation of CO leads to a complex mixture of
hydrocarbons which requires costly separations. Typically,
linear paraffins, olefins and some alcohols result with
relatively few branched products. In addition to the major
reaction pathways, side reactions can yield less desirable
products and affect activity and selectivity. These reac-
tions are summarized in Table 2. The water gas shift
reaction.(rxn. 6) can be very important on some metal cata-
lysts, as water produced in FT may react with CO forming
H2, which will shift the Hz/CO ratio in the reactor.

Fischer-Tropsch involves a great number of process

Table 2.

22

Adapted from references 75,70.

Important Synthesis Gas Reactions.

 

 

 

hcallmol
A o A °
paraffins
methanation 3H2 + CO = CH4 + H20 (1) ~23 -5l.3
02+ production (2n + 1)H2 + nCO = (2) -7 -39.4
Cunznez + nnzo
olefins 2nH2 + nCO = CnHzn + nHaO (3) -8
alcohols
MeOH 2H2 + CO = CHsOH (4) +4.5
02+ 2BR: 4' nCO = Cmnzneion + (5) ‘7
(n-l)HzO
water-gas shift CO + H20 = CO: + H2 (6) -5 -9.5
coke deposition C0 + H2 = C + H20 (7) -15 -31.9
nC = Ca
CO disproport— ZCO = C + CO: (8) -20 -4l.5
ionation nC = Ca
bulk carbide x! + C = NxC (9)
bulk oxide yM + O = lyO (10)
Figure 5. Influence of Process Variables on Fischer-Tropsch.

 

J Saturated Hydrocarbons

l<1

 

 

 

 

 

 

 

 

 

 

Increase of Pressure J .
U a
a
0 Increase of Conversion Rate 8
a - .3
8 Increase of the Content of Inert. A:
k «I
1. Increase of Temperature 2
.2 < , a
:3 Increase of Catalyst Concentration .8
0 Dr ‘
‘ Increase of CO Content ‘
3 U
3 Increase of the Gas Veloctity ] a

[I |

 

 

 

lpLUnsaturated Hydrocarbons J‘J

Reproduced from reference 189.

23

variables that affect the product distributions obtained.
Figure 5 shows a simplistic scheme that emphasizes the
influence of variables on product saturation and boiling
point.

The activity and relative proportions of product
groups also varies considerably with catalytic metal. The
specific activities of metals for the methanation of CO
were determined by Vannice (72) in terms of turnover fre-
quencies (molecules converted site'ls'l). Nickel, for
example, has a high activity and is one of the few metals
that affords close to 100% product selectivity in CO hydro-
genation. .Although that selectivity is for methane, meth-
anation converts synthesis gas of low heating value (5-6
M*Jou1es/m3) to natural gas with heating values of 35-40
M*Joules/m3 (73).

Iron is far more efficient at producing higher hydro-
carbons, including large proportions of alcohols and ole-
fins, although elevated pressures are typically required.
Iron catalysts however, are also good catalysts for the
water gas shift reaction, and exhibit a wide variety of
carbides under reaction conditions (74). Cobalt catalysts
on the other hand do not form carbides appreciably and give
low alcohol yields.

Although ruthenium suffers the disadvantage of higher
cost, it is also the most active metal for CO hydrogen-
ation. It has the capability to be supported easily in
high dispersion, and produces high molecular weight

24

hydrocarbons at low temperatures and pressures (76). The
tendency to produce higher molecular weight hydrocarbons on
these metals follows

Ru > Fe > C0 > Rh > Ni > Ir > Pt > Pd

1. Mechanisms

The mechanism of FT still a topic of considerable
debate. Polemics regarding postulated reaction mechanisms
abound in the literature and are summarized.in several
recent reviews (74, 76-78).

Many recent studies indicate that nonoxygenated inter-
mediates play a dominant role in the mechanism. This sup-
ports the hypothesis that dissociative adsorption of CO
occurs on the active metal surface. The adsorbed carbon
atoms are further hydrogenated to form methylene groups,
which can then propagate to form higher molecular weight
products. 'This mechanism, known as the surface carbide
mechanism (79), is shown in Figure 6A. It readily explains
many aspects of FT, but fails to account for the formation
of large quantities of alcohols and other oxygenates on a
metal such as Fe. The mechanism proposed by Storch,
Coulumbic, and Anderson (70) in fact predicts high oxy-
genate production by arguing for the condensation of sur-
face oxymethylene species, as shown in Figure 68. Finally,
the CO insertion mechanism, particularly advocated by
Pichler and Schulz (80), is the third mechanism recognized

by many researchers,as illustrated in Figure 6C.

25

z . o.= . ass: ... N= a
o
as /m
2 + o... + :8 F373 2
_
o
o\./n
a
on: + z + s ... «a d + fl
P u
/ \
x. «no esp/n ex /:

3

I + a n as: n

e = . so: = a z . n=2. .9. ~= :

_

=O\b/d 2:9

2 + cum + vac no. a: a

_ z + one ... u: x

o _

sex /= .26

.mdmmNMMdum nqnamuquua
T

x + z 8 + 3

cu: + z + z .Jdddquu z + z _ _ _
"—7 _ _ so/ «so :5

=e\ one soxc/n sexy/s «so

semus .so _dddddflduulflddu.

z + as: as. ~= q

o

 

0‘ 0<

.mma oosououon aonu panacea

 

 

.uluwssaoo: semaosom Acumenalhmmwumb .m osaumh

26

2. Nonselective Product Distributions

Widespread commercial application of the FT conversion
of synthesis gas (CO and H2) however, is severely limited
by the low selectivity obtainable for a hydrocarbon frac-
tion. Product distributions are typically broad and opti-
mization is hindered by an interdependence among selectiv-
ities. The only commercial application of FT processes is
in South Africa where the process is economical due to
inexpensive coal reserves. The interdependence among pro-
duct selectivities has been thoroughly discussed, using a
variety of SASOL iron based catalyst studies by Dry (81).

The product distributions obtained are governed by the
statistics of a linear, stepwise addition of single carbon
units to the growing hydrocarbon chain. .A competition is
present in which hydrocarbons can continue to grow on the
metal surface or undergo chain termination by desorption
from the surface. Assuming that the reactivity of interme-
diates is independent of size and that the relative proba-
bilities of continued C-C bond formation and product de-
sorption remain constant, a statistical model can be
derived and used to predict distributions and selectiv-
ities (70,82). This mathematical formulation is identical
to that derived earlier by Flory (83) and Schulz (84) for
linear polymerization processes. According to the Schulz-
Flory (SF) relationship the weight fraction of product with

n carbon atoms, termed Wn is

(n-l)
Wn = n C! (1" a: )2 (1)

27

Here (x is the probability of growth. If this SF distri-
bution is followed then a plot of log Wn/n versus n will
yield a straight line with a slope of log 0'. Schulz-
Flory product distributions have been a dominant feature of
FT since the early experiments. Yet, the consequence of a
product distribution following SF is that once process
variables are fixed the entire broad, product distribution
is determined. This inherent lack of product selectivity

prohibits commercial utilization of FT.

3. Selective Fischer-Tropsch

In a recent review, King (85) summarized current
approaches to more selective systems for the conversion of
synthesis gas to fuels and chemicals. 'To this I would add
the concept of structure sensitivity, so the areas are:

a. Limitation by pore size

b. Metal size dependence: structure sensitivity

c. Bifunctional catalysts

d. Alternative mechanisms
a. Limitation by pore size

The approach here is to confine the Fischer-Tropsch
active site within a small pore and force termination of
chain growth due to spatial restrictions. Production of
hydrocarbons longer than the pore limits would be severely
diminished in this case, and the product distribution would
shift to lower molecular weight products.

A number of brief communications have shown that with

zeolites (86-90), and aluminas with various pore sizes

28

(91,92), unusual product selectivities can occur in FT.

For example, by carefully encapsulating iron in the micro-
pores of zeolite Y (89), a catalyst was obtained that
selectively produced butenes (47%) and few higher hydro-
carbons (250°C). When a similar reaction was run at 300°C
methane production became dominant at the expense of butene
production. Further, when a catalyst prepared with large
iron clusters present on the zeolite external surface was
examined at 250°C, it gave a distribution that was similar
to that at 300°C for the encapsulated iron. The authors
suggested that the catalyst contained Fe active sites
within zeolite cavities initially but higher temperatures
induced migration and sintering of iron atoms which result-
ed in aggregates on the zeolite exterior. A similarly 4
prepared cobalt catalyst demonstrated.butene selectivity
near 70%, and showed resistance to Co migration up to
260°C. In all cases hydrocarbon production decreases
greatly beyond C 5.

Research with Co encapsulated A and Y zeolites showed
propylene as the only detectable product, and gave 70%
production of C4-C7 hydrocarbons respectively (86). .A Co/A
sample that had been reduced with H2 however, showed a
typical product distribution (about 65% CH4), which indi-
cated Co migration from the zeolite cavities.

Preparation of such catalysts by heterogenizing metal
cluster carbonyls has received considerable attention re-

cently (93), as it offers zero metal oxidation states and

29

intimate contact between metals in the case of mixed metal
complexes (94). The use of a metal carbonyl precursor also
eliminates the need for a reduction step prior to catal-
ysis, allows metal placement in zeolite micropores as many
clusters are < 7A, and permits infrared characterization of
the metal cluster in fig. The elimination of the reduc-
tion step is important as metals which may have been care-
fully encapsulated in micropores may migrate during reduc-

tion in hydrogen.

b. Metal particle size dependence: structure sensitivity

The role of pore constraints in shaping the FT distri-
butions however, is difficult to separate from that of the
unusually small metal particle sizes present within zeolite
cavities. When the small metal clusters used to introduce
metals into micropore cavities are exposed to elevated
reaction temperatures they decompose and can aggregate to
form ensembles of metal atoms (typically 1 nanometer'(nm)
in diameter). As activities and selectivities of such
small metal ensembles may differ from that of bulk.metal, a
metal size dependence seems plausible.

Based on this line of reasoning, and the non Schulz-
Flory distributions mentioned above, Nijs and Jacobs (95)
proposed an "Extended Schulz-Flory" model (ESF). Two
assumptions are central to this ESF model : (1) a metal
particle size dependence exists in FT which results from

the metal particle size and geometry imposing a maximum.on

30

the hydrocarbon chain length which can be produced, (2) on
each metal particle hydrocarbon chain growth can be de-
scribed by SF with the number of polymerization steps being
governed by assumption (IL. These assumptions imply how-
ever, that reaction mechanisms involving a one by one
insertion of C1 units into the growing M-alkyl chain must
be ruled out in favor of those having flatly adsorbed alkyl
chains.

To obtain theoretical product distributions based on
these assumptions Nijs and Jacobs (95) used Schulz-Flory to
calculate product distributions for all hydrocarbons less
than the maximum hydrocarbon chain length. The weight of
the hydrocarbon of maximum chain length (N) producible on
the metal particle of limited dimensions is

N-l n-l (2)
wu/N = 1— '2 a (1- a )2
1

Since the maximum length of the hydrocarbon is strictly
limited by the particle size in this theory, (N) could be
determined for each metal particle of diameter d by

d - ADN (3)

where DN is the particle diameter in terms of the carbons
in ON and A is a proportionality factor. For example, if
decane would just stretch across a metal particle of diame-
ter d (and if A were 1) ON would be 10. A skewed Gaussian
relation was used to calculate metal particle size distri-
butions, and d values were generated by equation 3.

To test the validity of this model non-SF distributions

31

in the literature were numerically fit by optimizing a ,
along with d, and the particle distribution variance from
the Gaussian curve. Figure 7 shows the reference FT dis-
tribution determined by Vanhove e_t; a; (91) for Co/alumina
compared with the product distribution generated by ESF.
The UN values used for A and B were 4 and 17 respectively
(a=0.89). ESF produced curves that provided a good fit
with the experimentally determined non-SF distributions.

The success of ESF in fitting non-SF distributions for
metals dispersed in zeolites or amorphous supports, implies
that to obtain selective FT distributions very narrow metal
particle size distributions, and high chain growth prob-
abilities are required. Broad particle distributions with
numerous different DN's would result in broad product dis-
tributions, and without a high probability of forming long
chain hydrocarbons the cutoff in production of higher
molecular weight products would not be detectible. One
great advantage of zeolites (and perhaps pillared clays) as
FT supports then may be their ability to retain narrow
metal size distributions and small metal sizes, as sinter-
ing of metals could be greatly diminished in the micro-
porous channel networks.

The distributions that ESF fit however, were from very
brief communications which lacked detailed product anal-
yses, kinetic data, or experimentally determined metal
particle size distributions. More rigorous investigations

of the relationship between product distributions and metal

32

 

 

 

 

 

 

 

Figure 7. (a) comparison between a experimental non-Schulz-
Flory distribution (open symbols,91), with an Extended
Schulz-Flory curve (closed symbols,95). Curve (A) is from
22 Co/Alea with mean pore radius 6.5nm, (B) 23 Co/Alea,
30nm mean pore radius. (b) particle size distribution (Poe)
for particle diameters used to generate the theoretical
Extended Schulz-Flory plot. Reproduced from ref. 95.

33

particle size have not substantiated the ESF model and it
has not been widely accepted. For example, Kellner and
Bell (96) showed product distributions of highly dispersed
Ru on alumina to be essentially invariant with Ru disper-
sion (inversely related to metal size).

Dependence of FT product distribution on metal size
falls within the scope of structure sensitive reactions.
In structure sensitive reactions the characteristics of
heterogeneous catalytic reactions are affected by the sur-
face geometry, or nature of a supported metal. As the
number of atoms in the particle decreases the proportion of
atoms in unusual crystallographic environments (ifia_corner
atoms) should increase offering various propensities for
unusual chemical reaction (97). Further, as metal part-
icles decrease below approximately 20 i in diameter (about
500 atoms) electronic properties begin to differ from those
of bulk metals (97,74,98,99). That FT is a structure
sensitive reaction is suggested by its similarity to
ammonia synthesis and.hydrogenolysis which.are known to be
structure sensitive, and the probability that a relatively

large site is required for the many components of chain

growth (75).

c. Bifunctional catalysts
This approach involves the production of primary or
low molecular weight intermediates (e.g. olefins), and

their conversion to higher molecular weight products on a

34

second catalytic site gig a chemistry different from FT.
Researchers at Mobil (loo-102) used ZSM-S to directly
convert alcohols and other hydrocarbon intermediates to
light hydrocarbons. The zeolite acid sites perform the
second transformation, allowing improved control over

product selectivites.

d. Alternative mechanisms

Finally, it should be noted that catalysts are availa-
ble which are not limited by the non-selective mechanisms
of Fischer-Tropsch. Two such processes are the use of
methanol as a feedstock, and the direct conversion of

synthesis gas by organometallic clusters (85).

D. Research Objectives

The diversity of cations used to permanently expand
layered silicate clay minerals has increased dramatically
since polycations of Al and Zr were intercalated into clay
interlayers. Aluminum and zirconium pillared clays have
high thermal stability and are active in the catalytic
transformation of organic molecules at elevated tempera-
tures, which has encouraged the synthesis of new robust
inorganic polycation pillars. The intracrystalline acidity
of these materials is a dominant feature in these catalytic
reactions. Yet, their application in high temperature
reactions, such as catalytic cracking, has been limited by
their high coking rates and low hydrothermal stability

relative to their zeolite counterparts. It has been of

35

particular interest then to demonstrate the catalytic
potential of these materials at moderate temperatures and
examine catalytic features of these materials other than
the intracrystalline acidity.

The primary objective of the present work is to inves-
tigate the catalytic potential and characteristics of cer-
tain pillared clays in Fischer-Tropsch, a moderate-temper—
ature catalytic reaction. Fischer-Tropsch requires that
catalytically active metals be dispersed in the pillared
clays“ The first method used to disperse metals in the
pillared clay galleries involves iron pillared clays, dis-
covered by M.S. Tzou (58). The catalytic research on these
iron pillared clays focuses on examination of: the activity
of these pillars in Fischer-Tropsch; hydrocarbon distrib-
utions obtained; and the stability of the pillars during
catalysis. The micropore location and chemical environment
provided by the pillared clay for the Fischer-Tropsch
active site is of particular interest as it may influence
the product distributions obtained.

The second method of highly dispersing Fischer-Tropsch
active metals takes advantage of the intracrystalline
acidity of alumina pillared clays to protonate ruthenium
carbonyl clusters within the clay galleries. This unique
synthetic method, was also discovered at Michigan State
University (103). We viewed this method as a means of
selectively introducing a Fischer-Tropsch active site into

the micropore regions of one of the most stable pillared

36

clays known, IRuthenium has a propensity for relatively
high molecular weight hydrocarbon production in Fischer-
Tropsch, and is one of the most catalytically active metals
in this reaction. .Another major focus of the present work
then is to investigate the activity, hydrocarbon distrib-
ution, and disposition of the Fischer—Tropsch active sites
formed by ruthenium cluster introduction into alumina

pillared clay micropores.

CHAPTER II

Experimental Methods

A. Clay Preparation

The Wyoming montmorillonite used in the preparation of
iron pillared clays was dispersed in distilled water and
then centrifuged at 2000 rpm for 15 min. to remove large
particles and carbonate impurities. The decantate was used
in pillaring reactions. ‘Arizona montmorillonite was used
for preparation of alumina pillared clays that were to be
loaded with Ru, as it contained less free iron oxides and
had little constitutional iron. Before pillaring however,
the Ca2+-montmorillonite from Apache County, Arizona (ob-
tained from the Source Minerals Repository, 1981), was con-
verted to the sodium exchanged form by stirring the dis-
persed clay in 1M NaCl overnight“ The clay was then di-
alyzed against deionized H20 until Cl' free, then sedimen-
ted 12 hrs in large graduated cylinders. All but the
bottom 200 ml was removed by suction. The clay suspension
was then centrifuged and the resulting solid treated with
150 ml 1N Na Acetate buffered to pH 5 by acetic acid, per 5
gram portion of clay, to remove the carbonates present
naturally in this clay; The mixture was digested at 70-

80°C for 3 hrs with occasional stirring with a rubber

37

38

policeman-fitted glass stir rod. The sample was trans-
ferred to a beaker, cooled, centrifuged, and washed one
time with distilled H20. Forty ml of 0.3M Na-citrate and 5
ml of 1N NaHC03 was added for each 5 g of clay remaining,
and the solution was warmed to 75-80°C on a hot plate. At
this point 1 g of Na28204 was added /5 g clay to the
stirred solution and the temperature carefully kept within
this range for 30 min so that free iron oxides could be
removed. When the solution had cooled it was again centri-
fuged and washed with H20. At this point the solution was
dialyzed again and resedimented, before distilled water was

added to bring the solution to 1 wt%.

B. Preparation of Alumina Pillared Montmorillonite (APM)
The synthetic method used to prepare the alumina pil-
lared clays used in this work was similar to that discussed
by Landau (104). The pillaring reagent used here was 50 %
w/w chlorhydrol solution obtained from Reheis Chemical Com-
pany. The amount of chlorhydrol solution required for a
given synthesis was weighed (2.6 g chlorhydrol/ g clay) and
diluted immediately prior to the pillaring reaction (50 ml
of distilled water/ g clay). The resulting solution (0.23M
in Al) was added slowly to a rapidly stirred 1 wt% clay
solution (15.5 mmol chlorhydrol/ mmol clay). The solution
was stirred for 2 hrs following completion of chlorhydrol
addition, and then the product was washed and centrifuged
repeatedly with deionized water to remove excess electro-

lyte. The alumina pillared clays with the highest basal

39

spacings and surface areas were obtained by carefully mini-
mizing the time that the pillared material was exposed to
large quantities of water. .After the first centrifugation
of pillared clay solution the mother liquid was decanted
and additional clay suspension was added to the original
centrifuge tubes without redispersal of the clay. This
process was continued until all of the original clay sus-
pension was centrifuged, The clay was transferred to an
Erlenmeyer flask with.a:minimum.amount.of distilled water
and any clay clumps were broken up by repeated shaking of
the stoppered flask; The amount of water required to fill
the centrifuge tubes was then added and the resulting
solution was shaken vigorously before the nonstop process
of centrifuging and washing was continued, The floccu-
lated, pillared clay obtained after 10-11 washings was air
dried on glass sheets in an area of good ventilation. The
material was scraped from the glass soon after drying was

complete and calcined for 3 hrs at 350°C under vacuum.

C. Immobilization.of Ruthenium Carbonyl on APM

The alumina pillared montmorillonite (APM) was reacti—
vated at 350°C for 4 hrs in vacuum in preparation for
ruthenium cluster immobilization. Ru3(C0)12 was added to
the cooled, activated clay (13 mg Ru3(C0)12/ 0.5 g clay)
using appropriate techniques for air/water sensitive com-
pounds. IMethylene chloride was then added and the solution

was stirred for 16-20 hrs according to the method of

40

Giannelis (104). The solution was then transferred to a
glove box for filtration and carefully washed to remove
excess ruthenium cluster, as physically bound clusters
would decompose and enhance Ru sintering during catalysis.
The green powder obtained was kept dessicated until used in
initial experiments, but as the material eventually changed
so that it was not catalytically active (about 2-3 months),
the material was later stored in the glove box and small
portions were removed for various studies.
D. Catalysis

Iron pillared clays were characterized for Fischer-
Tropsch catalysis in a 304 stainless steel (s.s.) fixed bed
flow reactor having a 7 mm inside diameter (Figure 8). Gas
flows were regulated using a Brooks model 5890 mass flow
controller. Approximately 0.3 g of Fe/PILC catalyst was
mixed with enough low surface area 8 -A1203 diluent (Norton
Chemicals, 10-14 mesh) to make a bed height of 7 cm. The
catalyst was held in the middle of the reactor tube by
plugs of glass wool. A 304 s.s. thermocouple (Omega) was
placed in the middle of the bed so that the oven tempera-
ture could be regulated by a Eurotherm model 990 tempera-
ture controller. Catalysts were dried in He flow at a gas
hourly space velocity (GHSV) of 300 hr '1 (V/V/hr, 35
ml/min) for 18 hours. They were then cooled to 25°C, the
gas was changed to H2 and the GHSV was increased to 428 hr-
1 before reduction in hydrogen began (200°C, 30 min; 400°C,

16 hrs). After the reactor had cooled, the flow gas was

41

 

 

.oa—axoa gas: noses». ans—cane

 

 

Umv

 

mp

you asaos soaosoc .c Gasman

.Ih

 

‘\
mi

 

 

 

 

\*\ \ \ \

 

 

 

 

\\\\U\

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

42

nauseous
sensuouslohao ssw
houéasa new

opus» unadulss sum
sonuouasoo Bonn sssl
opus» sash»:

souaum sosuml n.°

:ofiuoo—os asw\ opus» anon anon

.mn
.mn
.vn
.m—
.Nn
.um
.On
.m

housnswou ousnnoaaxosa
ash» suosvoua ennussovsoo
to; amhususo

mama goo: assuw

owns? «salsa:
ouasoOOILosu

chasm ousnsohm

h0d dchflflOU Ohm—flflhghlou— vgflflhflh

 

.hum a shaman

.m
.h
.w
.m
.G
.n

.N

43

switched to a premixed feed gas (Air Products, Hz/Co =2),
the GHSV was lowered to 300 hr'l, and temperature and
pressure were increased to 275°C and 120 psi. Condensable
liquids were trapped in the elbow of a steel vessel cooled
to -78°C (C02(s)/Acetone). Gaseous fractions were sampled
using a valved gas tight syringe (Supelco). Reactor con-
nections were made using Swagelok§>fittings.

Hydrocarbon fractions were analyzed on a F and M
Scientific model 402 by flame ionization, fitted with a 5'
x 1/8" stainless steel Poropak Q (Waters Assoc.) column.

Permanent gases were detected by thermal conductivi-
ty on a Varian model 920 GC with 6' x 1/8" s.s. Carbosieve
S-II (Supelco) column. Alkane and alkene standards (in He)
from Scott Specialty Gases were used for GC calibration.

Three major changes were made to this system for the
catalytic characterization of ruthenium loaded alumina pil-
lared montmorillonite (Ru/APM). In response to the low
conversions of Ru/APM at 18-20 psi, and sensitivity to feed
gas impurities (Appendix B), all technical grade gases were
purified by passage through a trap filled with SA molecular
sieve and cooled to -72°C (C02(s)/Et0H). Second, the reac-
tor described above was replaced with a 7 mm i.d. quartz
tube fitted with a glass to metal transition and flexible
s.s. tubing. The catalyst was placed atop a quartz frit in
the middle of the glass tube and catalyst and diluent
loadings were as above.

The elbow collection vessel used above was removed for

44

these studies at 18-20 psi and gaseous products were sam-
pled just below the reactor by syringe at a sampling port.
Condensable products were trapped at -72°C immediately fol-
lowing the heated back pressure regulator; The molecular
sieve trap effectively eliminated the formation of a mir-
ror, caused by impurity decomposition products inside the
quartz reactor.

Finally, the hydrocarbons were analyzed on a Hewlett
Packard 5890 GC, fitted with a 0a250 um x 60 m SPB-l capil-
lary column with 0.25 um film thickness (Supelco), by flame
ionization. This GC was also equipped with a thermal con-
ductivity detector, and permanent gases were sampled by an
in line GC switching valve and separated on the Carbosieve
S-II column.

The catalytic characterization of Ru/APM at elevated
pressures, required a redesign of the reactor system and
complete remoVal of background activity from the reactor.
The redesigned system is sketched in Figure 9. In this
system ultrahigh purity C0 (99.9%), H2(99.999%) and
He (99.999%) obtained from Matheson were further purified
over a Mn/Sioz catalyst (synthetic procedure, Appendix C),
that reportedly removed 02 to less than 1 ppb (105) and
serves as its own indicator due to a green to black color
change with.oxidation. 'The C0 purchased was contained in
an aluminum cylinder. Each gas was then passed through a 5
.A molecular sieve for H20 removal. These purifications

were accomplished in 1 liter 316 s.s. vessels that acted as

45

.:L<\su ho somuswuhansasao cuahnsaso Lou lunch. houosoa

 

 

 

 

2.23.3 Ar

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m—

 

 

 

 

 

 

 

 

 

 

.N

0—

 

% shaman

 

 

 

 

lasso... on.

 

 

 

 

46

down nauseoua amaMmsovsoo

whoa unsaniss tease:

boas—such ohsmnohaxosa veuooa

OuasoOOIHOAu

can ash~¢u¢0

~oo3 sasmw vowmssmus

vase fienhoa

guess“ wuasso

acansu sonmoa .nvssuw no»

so-osuso0 assassOAIOu oosshsm

.uN

.ON

.mm

.m~
.hn
.mu
.m~
.v~
.M~
.Nu

oossssh

no«~< Baa: sesame
“census anus cuoo

Gusto! oosmaOusm
seauosasoo 30mm soul

ebusb ads;

o>~s> asolo~a
\Jsmv ousaass

hennohusou Iona Incl
uoauuu sebum! m.°
chasm ousnsoha

o>~s> msmus~swos
scanshos ousasmol

hOAAShos sowhxo

 

hon .m vacuum

.u~

.e~

.a

.b
.m
.m
.v
.n

.N

47

reservoirs for purified gas. Purified gas flows were con-
trolled by Porter Instruments mass flow controllers and a
CM-4 control module. Before entering the reactor, gases
were additionally passed through a cold trap filled with
the low surface area alumina mentioned earlier at -72°C.
The reactor tube itself consisted of a 4 mm I.D. quartz
insert that was wrapped with teflon tape to make a light
seal with its 316 s.s. retainer. The entire reactor tube
was attached to the reactor system using teflon gaskets and
VCR fittings obtained from Cajongz Ru/APM catalysts were
placed atop a short plug of silanized glass wool that sat
on a quartz frit in the reactor insert, and held in place
by another glass wool plug. Typically 0.350 g of Ru/APM
catalyst was loaded in the insert, giving bed heights near
45 mm, and A1203 dilutents were not used. This reactor
system was used for all catalytic characterizations of
Ru/APM other than that shown in Table 8.

Although pretreatment conditions were varied (Chapter
IV), all catalysts were dried in.H2 at low temperatures
(100°C) before catalysis. Prereduced catalysts were loaded
and reduced at 1 atm in H2 as the temperature was slowly
increased stepwise to the maximum reduction temperature
(400°C). Catalysts were then cooled to the desired reac-
tion temperature in H2, and then the pressure was increased
(if desired) using the back pressure regulator. C0 flow
and pressure were equilibrated to the values required for

catalytic study and then added to the reactor by use of a

48

ball valve (Figure 9).

E. Electron Microscopy

Preparation of specimens for Scanning Electron Micros-
copy (SEM) utilized short segments of wood applicator
sticks, that had been flattened on one side and adhered to
the surface of a SEM stub. Clay particles that had been
air dried on glass and removed by scraping with a razor
blade, were wedged between these two wood pieces. ‘A thick
liquid agent (Tube Coat) was applied around the base and
exterior of the clay and atop the wood segments to enhance
conduction and eliminate sample charging in the SEM. A
light coat of gold was evaporated onto all surfaces using a
sputter coater. SEM was done on a JEOL JSM35-C that had a
Kevex Energy Dispersive x-ray (EDS) detector and Tracor
Northern digital beam control. EDS spectra linescans were
obtained by doing a computer controlled scan of the elec-
tron beam across the sample.

Suitable specimens for transmission electron microsco-
py (TEM) were obtained by subjecting the catalyst to soni-
fication in isopropyl alcohol for 5 min. The suspension
then was allowed to settle for 3 min and the upper portion
of the suspension was withdrawn. Holey carbon films (186)
mounted on copper grids were dipped 3-5 times in this
portion, which yielded numerous thin clay particles trapped
on the grid. The Philips 300 TEM, and the SEM mentioned
above, are located at The Center for Electron Optics at

Michigan State University.

49

Post catalysis Ru/APM specimens for scanning transmis-
sion electron microscopy (STEM) were prepared in a fashion
similar to TEM samples.

Most of the results for Ru/APM were obtained on a
Vacuum Generators Ltd. HB501 STEM equipped with field-
emission gun, Link Systems spectrometer for x-ray microa-
nalysis (atomic number 11 and higher), electron spectrome-
ter, annular dark field detector, and liquid nitrogen
stage. The instrument is part of the MSU HREM facility.

On the basis of the virtual objective aperture size and
condenser lens currents used for our studies an electron
probe diameter of 1-2 nm and a beam current of 1-3 nanoamps
are expected. X-ray microanalysis was performed using
collection times of 100 seconds. Count rates were low when
the sample was mounted on a fixed stage, but the use of a
tilting specimen holder inclined 20° from horizontal to-
wards the detector greatly improved the counting statis-
tics. EDS analysis was limited to thin areas of the speci-
mens, so that corrections for atomic number, x-ray ab-
sorption or fluorescence were not required.

The method of combining the inelastic electron signal
for electrons with small energy losses (those deflected
through small angles, bright field detector) with the elas-
tic signal for electrons with large energy losses (large
angle deflection, dark field detector) was also employed
for Ru/APM STEM studies. The procedure applied in this

work for this technique, called z-contrast imaging, was

50

similar to that described by Treacy e_t 51; (168). The
inelastic signal (for energy losses 15 eV<AE<30 eV) was
electronically combined with the elastic signal (50 um
objective aperture used) to minimize the contribution of
atomic scattering of APM in the dark field image. By
subtracting the inelastic signal from the elastic signal,
or taking the ratio of these signals, the contribution of
the heavier ruthenium microcrystallites in the image could

thus be enhanced relative to the clay support.

F. Physical Measurements

Surface area measurements for iron pillared clays were
determined on a Perkin-Elmer-Shell model 212 sorbtometer.
Alumina pillared clays were characterized for surface area
on a Quantachrome Jr. sorptometer. Measurements on both
sorptometers was done at -196°C using nitrogen as a sorbate
and helium as a carrier gas. The gases used with the Quan-
tachrome were ultrahigh purity (Matheson, 99.999%) whereas
the 212 used technical grade gases purified with a liquid
nitrogen trap. Samples were outgassed at 350°C under
flowing Ar for 2 hrs with Fe/PILC, but outgassed at 350°C
under vacuum for APM measurements. In both cases measure-
ments were taken at three partial pressures of nitrogen and
surface areas calculated by the BET theory (187).

The Quantachrome Jr. was also used to determine de-
sorption isotherms from which pore size distributions of

APM could be calculated. The values of P/Po (0.98 to 0.05)

51

were obtained by varying the ratios of flow rates of He and
N2 using a linear mass flow controller system purchased
from Quantachrome. The calculations used were those found
in the Quantachrome Jr; manual which follow the method
described in greater detail by Lowell and Shields (188).

A Philips x-ray diffractometer (Ni filtered CuK) was
used to determine the d001 reflections of pillared clays.
Samples were prepared by air drying 1 ml of 1 wt% clay
suspension on a glass slide. The slide was then calcined
along with the bulk pillared clay for 3 hours. Iron pil-
lared clay samples were calcined in air whereas APM was
calcined in a vacuum (about 10'2 torr).

Adsorption of organic molecules was studied using a
McBain balance equipped with quartz glass springs. Samples
were activated at 350°C under dynamic vacuum for 2 hrs
prior to measurements.

Infrared spectra were recorded on a Perkin-Elmer model
475 grating spectrophotometer by mixing the samples with

mineral oil and placing between CsI disks.

G. Chemical Analysis

Elemental analysis of iron pillared clays was carried
out at the Department of Toxicology Department (inorganic
laboratory) at Michigan State, using a Jarrel-Ash 995 Atom-
Comp analyzer. J.T. Baker instra-grade Si, Al, Fe, Mg and
Na were used as instrument standards and NBS clay 98a
served as a clay standard. Clay samples ULOS g) were

fused with lithium borate “L3 g Gold label, Aldrich) for

52

2 min at 1000°C and the resulting glass quickly dissolved
in 3% HNO3(30 m1), before being diluted to 100 ml with
deionized water.

Ru/APM samples were sent to Galbraith Analytical Labo-

ratory, Inc. (Knoxville, TN) for Ru analysis.

Chapter III

Iron-Pillared Clays

A. Introduction

Iron has been one of the principal components in innu-
merable FT catalysts, yet iron alone supported by a sub-
stantial number of carriers has been very ineffective in FT
(106). Highly dispersed iron supported by oxides such as
A1203, Si02 and MgO is very difficult to reduce to Fe0 due
to strong metal support interactions. If these inter-
actions are diminished reduction to Fe0 can.be accom-
plished, but metal crystallites are obtained due to migra-
tion and agglomeration. This results in ineffective use of
metal atoms for catalysis and loss of the potentially
useful properties of small metal sites, mentioned in
chapter I. Optimum.iron-support interactions that result
in small, yet stable, zero—valent iron atoms have been
elusive for the most part. ‘Additionally, when iron is
supported by Si02 and A1203 the iron oxide can migrate even
before reduction, and metal crystallites continue to grow
after formation (106).

Brenner and Hucul (107) however, were able to obtain
high iron dispersions below 300°C on alumina using various

iron carbonyls.'Temperature-programmed desorption studies

53

54

suggested that different catalytic properties might be
achievable from clusters of different nuclearity. Hughes
£2.2L;(1°3) impregnated magnesia and alumina with iron
carbonyls in pentane and found that small iron metal and
oxide particles produced large yields of ethylene and
propene initially, but normal hydrocarbon production was
found within a few hours. Iron agglomeration may account
for the change in selectivity. Lung gt a}; (109) also
achieved success in highly dispersing Fe by using a variety
of carbon based supports that offered easily reducible
Fe3+. Carbon monoxide turnover numbers were eight times
higher with these catalysts than Fe/A1203.

Efforts to extend the family of hydroxy-metal pillared
clays by MLS. Tzou (58) may be of particular interest in
highly dispersing iron. It was shown that smectites could
be pillared with hydroxy-iron polymers. 'These microporous,
expanded clays had basal spacings as high as 30 A, rel-
atively high surface areas, and thermal stabilities near
500°C. It occurred to us that if a thin outer sheath of
the pillars could be reduced, small Fe sites could be
generated within the clay galleries that may be active in
FEB These active sites would be unique among FT active
sites as they would be very small and confined within
microporous cavities. 'The demonstration that pillars could
be catalytically active would also be unusual as catalytic
transformations involving pilared clays typically result
from intracrystalline acidity.

SS

1. Hydrolysis Chemistry of Fe(III)

The hydrolysis chemistry of Fe(III), used by Tzou for
Fe-pillar formation, has been the subject of a great number
of investigations due to its widespread technological im-
portance. The knowledge obtained in these studies is the
subject of comprehensive reviews (110,111), so only the
salient features will be presented here.

The initial species formed due to increased acidity
of water coordinated to metal ions and subsequent hydroly-
sis, are low molecular weight products. The addition of
base, or elevated temperatures, induces color changes in
Fe(III) solutions from yellow to orange to deep brown,
which are indicative of continuing hydrolysis through the
general reactions of olation and oxolation mentioned in
chapter I. Early studies by Spiro gt a}; (113) indicated
that polymers were formed with continued hydrolysis, and
that the polymer fraction in fresh Fe(III) solutions
increased linearly with OH/Fe ratio. The molecular weight
of the polymer was found to remain relatively constant in
the OH/Fe range of 0.5 to 2.0, but increases considerably
when the ratio was raised further. Murphy e_t_ 13:. (114-117)
showed that discrete, spherical polyoxocations were formed
initially in the OH/Fe range mentioned above. These
discrete spheres, observed by electron microscopy, were
Ins-3.0 A in diameter. With time spheres linked together
to form rods containing 2-6 spheres, then coalesced forming

rods 200 x 30A, which eventually aggregated into rafts of

S6

rods. These investigators reanalyzed the earlier work of
Spiro gt El;.(113) and calculated the molecular weight of
these spheres to be near 105. Assuming Fe(OI-I)2.5 as a
monomer formula Flynn (111) calculated that the spheres
contain roughly 100 iron atoms“ This process of polymer'
growth, coalescence, and aggregation is shown in Figure 10.
The initial stages of polymerization represented by spheres
and rods are however very far from thermodynamic equilib-
rium, and hydrolysis can continue for months or years until
final forms such as goethite (a'-FeO(OH))‘are formed.
Polymerization is accelerated by increasing parameters such
as: temperature (118), Fe(III) concentration, pH, ionic
strength (114), and by changing the anion (117). With
chloride ion, agglomeration of spheres was fast and rods
were formed after 3-4 hours. However, electron microscopy
showed only spheres at comparable aging times for nitrate
or perchlorate salts (114-117). Agglomeration of the rods
to rafts however, was slower with chloride salts. Thermo-
dynamic data for the formation of potential precipitates

are given in Table 3.

2. Iron Pillared Clays

The high basal spacings ( > 18A) for the iron pillared
clays synthesized by Tzou (58) were significantly larger
than those previously obtained for iron(III) smectite
intercalants (119). Tzou studied the effect of OH/Fe,

aging time, temperature and iron counter anion on iron

S7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-H* FeOH”
Fe3" —‘._ Fe2(OI-l)2‘*
1H+ +H‘ F9(0H)2*
(IOS) (I)
Preciplloled solids. *l-l+ -H+
e.g., Iepldocrocile (ID2 to l03) (IOZ)

 

 

 

Fresh polymer spheres

 

 

 

2 lo 4 nm
+H" *
«0410:05) '”
(I05)

 

 

’Hordened polymer spheres
2 lo 4 nm

 

 

 

-H"
005nn0fi

 

Aged polymer rods

 

 

 

-H*
(l0510l07)

 

Aged polymer rofls

 

 

 

-H’
(IO7 lo IO°l

 

Precipiloled solids.
e.g.. goethite

 

 

 

Figure 10. Iron(III) hydrolysis and
polymerization, as shown by Flynn (111).
The numbers in parenthesis refer to
relative reaction rates.

58

Table 3. Solubility Products of Fe(III) Oxides. Adapted
from 111, 25°C, Zero Ionic Strength.

 

1/2 FezOs(s) + 3/2 320 = Fe3* + 308‘

re0(oa)<.) + n20 ;=e Fe3+ + zon-

solid phase -long
(1/2) CI-Fezos (hematite) 41.7

FeO(OH) (goethite) 41.7

”Fe(OH)3” amorphous 37.1-39

 

Adapted from Flynn (111), 25°C, zero ionic strength.

Table 4. Sorption of Probe Molecules on Fe/PILC.

 

Tzoug58) Yamanaka et a1. 59
Probe Amt. Probe Amt.
Molecule K. D. Ads. Molecule K.D. Ads.
benzene 5.8 2.6 benzene 5.8 4.6
neopentane 6.2 1.7 p-xylene 7.3 3.2
1,3,5 TEB 9.2 1.2 mesitylene 8.4 2.4
PFTBA 10.2 .9

 

1,3,5 T83 = 1,3,5-triethylbenzene

PFTBA = perfluorotributylamine

K.D. = kinetic diameter (A)

Amt. Ads. = amount adsorbed (mmol/g) at 25°C

59

pillared clay formation.

In a series of syntheses with various OH/Fe ratios
(0.2M FeCl3 and Fe(NO3)3 ) the basal spacing of the air-
dried Fe-pillared clay (Fe/PILC) increased steadily in the
range OH/Fe-0.0 to 2.0. At OH/Fe=0 the d001 was only 12.33%
indicating the intercalation of only the small, low molecu-
lar weight species. With OH/Fe=l.0 a material was found
with d001=23.5A and at OH/Fe=2 the Fe/PILC exhibited a
d001=24.8A. 'With OH/Fe=2.5 however, interstratified clays
were obtained. Pillaring using solutions with OH/Fe from
1,0 to 2.0 then allows intercalation of iron-polycation
spheres that are similar to the dimensions found in the
previously mentioned studies on Fe(III) hydrolysis chemis-
try.

When smectites were pillared with FeCl3 solutions that
had aged for various times it was found that basal spacings
increased with longer aging times. This corresponded with
decreased pH values of the pillaring solutions with time.
Although basal spacings increased from 27.2A for clays pil
lared with solutions aged 3 hrs to 29.5A for solutions aged
7 days, surface areas decreased from 351 to 270 mz/g, re-
spectively; Elemental analysis showed the iron content per
unit cell increased from 648 to 8.8 Fe/cell for these same
clays. As Fe-polycations grow they decrease in overall
charge and the intercalation of more polycations to balance
the layer charge, is consistent with these trends. The

steady increase in basal spacing with time could be

60

attributable to a higher proportion of the larger spheres
seen by EM being intercalated. With aging times greater
than a day rafts and perhaps bundles of polymer rods are
intercalated which would dramatically decrease interlayer
surface areas.

The dependence of Fe-polycation formation on counter
anion was also examined with respect to the ability to
prepare Fe/PILCs. Although the chloride, perchlorate and
nitrate salts gave similar, well-expanded clays, sulfate
solutions gave low basal spacingsm This was in accordance
with anion penetration arguments which explain that sulfate
competes favorably with hydroxide for iron centers. This
inhibits Fe-polycation formation in the case of sulfate
Fe(III) solutions.

The Fe/PILCs prepared by Tzou retained high basal spa-
cings (>20A) up to near 500°C in air, decreasing from their
original spacings by only 1-2 A. Additionally, these mate-
rials had fairly large micropores as they absorbed 1¢2
millimole/g 1,3,5-triethy1benzene (9.211) and 0.96 milli-
mole/gram of perfluorotributyl-amine (10.4A).

Recently, Yamanaka gt g;t_(59) synthesized an iron-
pillared clay by the intercalation of acetato-hydroxo iron
(III) nitrate in montmorillonite. Subsequent calcination
of this material yields iron oxide pillars in the galleries
which yield a product with a 16:7A basal spacing and sur-
face area of 300 mz/g at 500°C. .As elemental analyses

showed the mole ratio of acetyl:Fe in the materials to be

61

less than that expected,based on stoichiometry, they sug-
gested that the acetate complex was intercalated in par-
tially hydrolyzed forms. If this is so, the gallery may be
filled with hydrolyzed iron species as is the case for $2
EiEE hydrolysis preparation methods discussed in chapter I.
Yet, the large surface area and high sorption capacities of
this material suggests that it is quite porous.

The nature of the pillaring reagent used by Tzou (58),
Yamanaka (59), and Oades (120) is considerably different.
The work reported here, and that of Tzou, involves the
intercalation of freshly formed or hardened polycation
spheres (and sometimes rods) based on the terminology of
Figure 10. The population of polycation sizes inter-
calated, and the gallery height obtained, depends on the
set of conditions chosen (pH, OH/Fe, age, etc.), which
determines the extent of Fe(III) hydrolysis. Polycations
in the initial hydrolysis stages gave the best combination
of gallery height, and surface area. Even though acetato-
iron complexes were intercalated by ion-exchange in the
work of Yamanaka g gL (59), after the Fe/clay complex was
held in suspension for 3 hours partially hydrolyzed forms
‘were intercalated also. .As mentioned the lack of control
over pillar hydrolysis and chloritic interlayer formation
are drawbacks of this method.

Oades (120) hydrolyzed Fe(III) nitrate solutions and
used NaOH to promote polycation formation. The addition of

such a strong base however encourages precipitation of the

62

iron salt. He used solutions of this salt aged 2 to 4 days
at room temperature that were filtered before being added
to clay solutions. His solutions undoubtedly contained
substantially more higher molecular weight polycations than
used by Tzou. Also, he simply added the polycations to
clays without washings and low gallery heights were
obtained for the iron clays (10.3 A) and Al-clays (14.7 A)
after heating to 150°C.

The basal spacing between 25 and 295. found by Tzou are
much larger than those found by Yamanaka gt _a_l_:_ Tzou's
materials had lower sorption capacities for large organic
molecules (Table 4), and thus may have more micropore char-

acter.

B. Synthesis and Physical Properties

A synthetic procedure similar to that of Tzou (58) was
followed in this work to obtain smectite clays pillared
‘with polynuclear iron oligomers.‘Although both the nitrate
and chloride iron salts were used here, clays pillared with
iron chloride gave sharper XRD patterns (indicating more
well ordered materials) and were the preferred pillar pre-
cursor. Although detailed synthetic procedures are covered
in chapter II, it is important to point out that the iron
to clay ratio in this work was nearly half that used by
Tzou (30mmo1 Fe3+/milliequivalent of clay 3g; 70).

The washing procedure used proved to be crucial in the
synthesis of a well-ordered iron-pillared clay. Several

processes may occur simultaneously during the washing

63

process including: alteration of pH, change in the charge
upon the clay layers, continued pillar hydrolysis and
growth, and removal of excess iron oligomer. Washing the
pillared clay with deionized H20 steadily increased the pH
of the clay as illustrated in Figure llA. An increase in
pH can influence the net charge on the clay surfaces and
the extent of hydrolysis (and charge) of surface bound
cations (120). This is turn affects the balance between
attractive and repulsive forces of clay sheets and deter-
mines whether individual clay platelets are dispersed or
aggregated together (flocculation) in clumps or tactoids.
x-ray diffraction patterns taken of the Fe/PILC after each
wash revealed a well-ordered, pillared clay with high
gallery heights (d001 > 14A) only after the ninth washing
(Fig 11B), which is also the point at which flocculation
occurred. Additional washings resulted in further
increases in layer ordering as shown by the sharper XRD
reflections. 4Attempts to bypass the laborious washings by
raising the pH to the point of flocculation with addition
of solid Na2C03 after a few washings, gave a Fe/PILC that
indeed was pillared but not well-ordered. After two or
three washings the pH altered Fe/PILC appeared just as
well-ordered as control Fe/PILCs having the same number of
washings.

The pH increase through washings, or Na2CO3 addition,
may result in a substantially increased degree of pillar

hydrolysis and growth within the clay galleries. If

64

252A
V
A
“lash
12
44
604% 11

 

 

 

 

N.

o 3 E '9 3 11 6

Number of washings Degrees 2 9

Figure 11. Change in the pH of Fe/PILC sols during
the process of washing (A), and XRD patterns (B) of
air dried Fe/PILC films after selected washings.

The arrow in (A) indicates the pH and washing at which
flocculation first occurred.

65

uncontrolled, this may result in a chloritic interlayer
similar to the t2 gttt_pillar preparation method. When a
portion of the Fe/PILC washed twelve times was dialyzed
against deionized water for one week the material showed a
broad hump at 14.2 A confirming chloritic interlayer form-
ation. Other experiments using dialysis of Fe/PILC at
various stages in the washing process showed that dialysis
did not lead to flocculation, perhaps because iron-olig—
omers could not pass through the dialysis membrane. Due to
the relatively facile hydrolysis of Fe/PILC, washings were
done as quickly and as continuously as possible. The
nonequilibrium state of intercalated Fe-polycations also
must be considered when deciding on the method of PILC
drying.

Oades (120) has demonstrated that the amount of metal-
polycation adsorbed on a clay can dramatically affect the
clay surface charge and the point at which flocculation
occurs" Thus, the removal of excess oligomer during wash-
ings can play a role along with the pH in controlling floc-
culation. IExcess Fe-oligomer may also block the PILC
micropore channels and form large Fe crystallites during
catalysism Thus, circumventing the numerous washings by
dialysis or base addition would not be beneficial. The
need to remove this excess though must be balanced with the
pillar hydrolysis that will undoubtedly occur with the time
and higher pH required for numerous washings.

Table 5 summarizes the relevant physical data for the

66

huhsfi OaGOHnws sch amsaa semsc canovouo.soaum0Aloo boas—Lsasm.o

sodssoasss undo am .ss: 9 awed

..uoeanuoooocooe_.t

«canouuu neeaesooooceou_ :ee.e .n:eoa> .ueehn ...: en toemusao.o

«soon :~.e .tee .ooeen

..es en toesosuo.e

«goon I~.°.hm¢.Oecmn .aaa N flosmuuso.u

 

 

 

 

can m.o s.o~ as.» an m 9.4m mm exchmmwu .o

ecu m.«~ m.~« ms.m en es 3Ammc eons .n

new e.m~ «.mm ma.o n on . she: uses .<

dw\~lvc0h< «not «one ounce auss\0h adshAme< ww~o flwl neonas<
canteen flee emu tosses .nos aes-

 

.oeee\oe tee eeae fineness; so seen-em .m e_eue

67

iron pillared clay used in this work with that of Tzou and
Yamanaka gt 1]; Tzou's Fe/PILC contained approximately 40%
more interlayer iron than the clay prepared in this work.
This is due to the shorter aging times used here and the
lower ratio of iron oligomer to clay; .A higher surface
area was obtained here however, which is consistent with
the trend observed by Tzou. The amount of iron in A is
also less than that intercalated by Yamanaka gt git, yet
surface areas are similaru The most significant feature of
this comparison is the much higher basal spacing for cal-
cined Fe/PILC A than that obtained by Yamanaka. It is also
somewhat higher than that obtained by Tzou under similar
iron concentration and pillar age. It is of interest that
C used only a fivefold excess of iron whereas both A and B
used larger excesses. Of course the Fe excess required the
extensive washing procedure, while Yamanaka gt gltlcentri-

fuged and washed "several times".

C. Fischer-Tropsch Catalysis

The Fe/PILC summarized in Table 5 was typical of well
ordered Fe/PILC, so the catalytic and physical features
presented in this chapter are those of this material.
Figure 12 shows SF plots (eqn. 1) of the product distribu-
tions obtained using the catalytic methods outlined in
chapter II, for three sampling times. Although the lines
drawn through the data points for all three times have

nearly the same intercept their slopes steadily increase

68

 

 

 

1.0:
‘ o
' E:
1

0.1g
1

a I

\ I

8

3 1

.01 .
: = 0.49
1 = 0.47
. ‘ = 0.42
. G
0 1 2 3 4 5 6

Carbon number (n)

Figure 12. Schulz-Flory plots of Fe/PILC hydrocarbon prod—
uction after 60 min.® , 103 min. A , and 1259 min. B of
reaction. Reaction conditions were 120 psi, 275°C, Hz/CO=2,
2100 hr"l GHSV, and 1.73 contact time.

69

with longer run time. The probabilities of chain growth
obtained from these slopes increase with time, reflecting
the trend towards higher hydrocarbon production. This
contrasts with the work of Krebs _e_t a_lt (121) who applied
surface analytical techniques along with catalytic measure-
ments to compare FT features of reduced and unreduced
magnetite (Fe304). In reduced magnetite and clean iron
foils the a -values steadily decreased from 6 to 200 min
which corresponded to the gradual increase in carbon depos-
ited on the sample. They suggested that the "clean",
reduced iron surface had the greatest ability to produce
high chain growth probabilities (<1 ).

The nature of the catalytic active site for iron is
still a matter of considerable debate as the potentially
active phases such as: iron oxide, reduced iron, and the
numerous iron carbides formed may all be components of
hydrocarbon production (122). The formation of carbide
phases is represented below.

Fe304 e=‘ Fe° ‘1‘ FeC (X, e, 5'...)
Several studies (123,124) support the proposal that as-iron
is active initially in FT, but rapidly deactivates due to
the formation of graphitic overlayers. Yet, others
(125,126) observed increasing activity with time which cor-
related with accumulation of iron carbides. Iron carbides
in FT have been discussed in detail elsewhere (106).
For all three data points, methane falls above the

weight fraction suggested by the least-squares line drawn

70

in Figure 12. This is particularly true at 60 min. This
deviation is typical of PT distributions as excess methane
may be formed from hydrogenolysis of higher hydrocarbons.
Another general trend observable in Figure 12 is that
deviations from the line decrease with time. For example,
C2 production is far lower than expected in the SF plot at
60 min, slightly less at 103 min, and on the line with the
other hydrocarbon components at 1259 min.

These deviations can be depicted in a more illus-
trative way by using the a value obtained for each sampling
time to produce a theoretical SF distribution by applying
equation 1. Figure 13 shows such a comparison at 60 min
between experimental and theoretical SF-derived data. The
relative magnitude of these deviations can be better appre-
ciated in this manner and the decrease in deviation with
higher hydrocarbon easily noticed. Comparison between the
experimental and SF theoretical distributions at 1259 min,
in Figure 14 however, shows that indeed with time devi-
ations from Schulz-Flory distributions become minor with
increasing time on stream (TOS).

The distribution of olefins and paraffins, conversion
data and the composition of the trapped reactor liquid are
summarized in Table 6. The initial methane production of
about 50 wt% drops considerably to 32% CH4 at 103 min,
which further indicates higher hydrocarbon production. The
increase in conversion with TOS may be accounted for by the

formation of a surface carbide vital to hydrocarbon

71

SCN

 

3Ch

"n

ZCM

EO

10< A

OD

 

lu1to

v '

1 2 3 4 5

Carbon number (n)

0,1

Figure 13. Experimental hydrocarbon production<3 for

Fe/PILC (60min. TOS) compared to calculated A (Schulz-
Flory) distribution. A chain growth probability of 0.42
was used for the SF plot. Experimental data from Fig. 12.

72

3(fi

ZCM

OD>

10'

C)

DC)

 

 

«‘1

1 i :3 4 5 6

Carbon number (n)

Figure 14. Experimental hydrocarbon production C) for

Fe/PILC (1259min. TOS) compared to calculated E](Schulz-
Flory) distribution. A chain growth probability of 0.49
was used for the SF plot. Experimental data from Fig. 12.

73

 

 

 

 

 

mote xv. .sosu um.~ .uoez xe.s .omu u~.mo "eesasuee-eo assess teasers

.-eseem so asl\~l mm .Nnoo\~=.moee-.oomsm be :eseoeoe

.nte e .ooeee .~= ea sconce:

m.~ e.m m.a s.s. m.n s._~_ s.s_ _._ s.sn sq. m._ ohms
e.. N.e e.m s.s. o.n m.h_ s.s. v.~ s.sn se. ~._ me.
e.~ N._ :.n s.s. n._ n.o~ m.e a. s.sm we. ms. eo

so no mo v0 loo "no Inc "No no so sowss0>=ou a AswlvaP

u saweoz
.eeeseeeeauee acetate as oenexos sausage so me seal-5m .o oases

74

production as mentioned earlier; The unreduced magnetite
studied by Krebs gt 3;; (121) and unreduced hematite
(Fe203) studied by Dictor and Bell (127) both showed
increased activities in the first 20 hrs of exposure to FT
reaction conditions. These results are consistent with the
above mentioned observations of the present Fe/PILC study.

Additionally, the values of arobtained in this work
are similar to that of Arcuri gt & (128) for Fe/Sioz
(a=0.43 at 14 atm, 250°C, H2:CO=3), and Krebs gt _a_l_. (121)
(1 bar, 298°C, szco=3) where a ranged from 0.46 to 0.39.
Dictor and Bell (127) showed arto increase with decreasing
hydrogen partial pressure (pHZ), increased pCO, and
decreased temperature whereas the gas flow rate had negli-
gible affects. Although the a values for these studies are
not directly comparable, they are very similar, suggesting
that Fe/PILC catalyzed hydrocarbon chain growth in FT is
similar to other dispersed iron catalysts.

The dominance of water in the trapped liquid (Table 6)
is typical of iron-based FT catalysts where this is the
primary path of oxygen removal from the catalyst surface.
At elevated pressures the water gas shift reaction
decreases (128). At the pressures used in Fe/PILC F-T, the
C02 produced typically accounted for 6% of the total CO
converted. The liquid fraction also contained significant
portions of methanol and ethanol and smaller amounts of
propanol. Methanol production increases with higher reac-

tion pressures also, and may become the dominant product

75

(second only to CH4) for iron and iron-cobalt catalysts
(128). Branched hydrocarbon production was below the
detection limits of the analysis system used for most
Fe/PILC studies.

In numerous catalytic runs with well-ordered Fe/PILC
the olefin/paraffin ratio decreased with increasing conver-
sion. This is shown in Figure 15. The decreasing values
of Cn‘/Cn' versus conversion agrees with previously
obtained results (128,130). It has been proposed that
ethylene plays a special role in the chain growth process
(71,130) as it can be readsorbed and incorporated into
growing hydrocarbon chains. Another suggestion is that
terminal olefins (in particular ethylene) are the principal
products in FT (127), and that they can undergo further
reactions to form paraffins and internal olefins.

Increased conversion then may result in increased incorpo-
ration of light olefins into growing hydrocarbon chains.
The incorporation of these olefins also explains Cz-C3
hydrocarbon production below that predicted by SF, such as
C2 in Figure 12. The olefin/ paraffin ratio typically
increases with decreasing Hz/CO ratio (olefin hydrogenation
requires H2), decreasing pressure and temperature, and
increasing gas velocities (122).

The fact that C2 deviations from SF decrease with
increasing time and conversion for Fe/PILC is interesting,
because the above argument suggests that deviations should

increase with larger conversions as more C23 is

76

 

144
O

124

10‘
In J
E: 8
"n
U 6-

G

.5 4«
z: N
s 2-
s
s
Q
\~ 41
s
c
.2
o

?'
C)
A

/C2’
[>

[>

OJV

 

 

0.0 (,0 2.0

Conversion(2)

Figure 15. Olefin/paraffin production for Fe/PILC
catalysts as a function of conversion for Ca‘/Ca'
and 023/02‘.

77

incorporated into growing chains. Also, the increase in a
with time is a curiosity. This may be due to the greater
propensity for light hydrocarbon production within the con-
fines of the Fe/PILC microporous network.initiallyu The
fact that a changes with time suggests that the active
site is changing with time, perhaps due to the changing
nature of the iron pillar.

Previous studies with precipitated iron (122, 131,
132) and hematite (127) catalysts have shown that more than
one a can be found in FT synthesis. In these studies,
hydrocarbon production beyond a certain carbon number
(typically C7_10 and higher) was significantly greater than
predicted by the SF line for lighter hydrocarbons, and two
chain growth probabilities were required to adequately fit
the distribution. The origin of the break in, a is not
well understood at present, but in many instances potassium
was used as a promoter. It has been suggested
(122,131,132) that two different active sites arise from
sites containing various levels of K5 Potassium is known
to promote the production of higher molecular weight hydro-
carbons, and sites of varying alkali concentration may
differ in their strengths of hydrocarbon adsorbtion ex-
plaining this effect. Yet, Huff and Satterfield (132)
showed that similar iron catalysts without alkali pro-
moters, could also generate product distributions requiring
two cr's. It appears that nonuniformities in structure

and/or composition of the active site can be responsible

78

for the bi- a anomaly. In the present work we can fit the
product distribution with one a value but the fact that (1
changes with time suggests that there is a tendency toward
two a values.

It is plausible that a structural or compositional
reorganization of the Fe/PILC may account for the unusual
shift in a with time, and the tendency toward higher hydro-
carbon production. To examine such possibilities the fun-
damental nature of Fe/PILC catalysts and the local ele-
mental composition was examined using scanning electron

microscopy.

D. Scanning Electron Microscopy

Electron microscopy has been invaluable in estab-
lishing the morphology, structure, classification, and
properties of clay minerals (15,133,134). Clays can
exhibit a wide variety of textures as a result of the
layers being weaved into ribbon-like, cylindrical, or
matted fabrics. ‘High resolution electron microscopy and
microanalysis have been used to discover new topologies of
multiple chain silicate structures, correlate local compo-
sition with polytype, and investigate atomic structure near
framework imperfections (135). In the case of montmoril-
lonite the morphology of the elementary layer is platey,
but owing to the flexibility of the layers, isomorphous
substitutions which induce curvature or undulation, and the
turbostratic nature of layer stacking an aggregated

particle rarely exhibits three dimensional order. Thus

79

the texture of montmorillonite is raglike, even though the
elementary layers are regularly stacked in the vertical
direction.

The edge-on SEM image in Na+-Wyoming montmorillonite
in Figure 16a demonstrates a number of elemental plates
clinging to the edge, the stacking and bending of numerous
layers (lower left), and the wavy morphology (Figures 16a
and b). Early EM investigations noted that the intercal-
ation of organic cations in clay galleries stiffened mont-
morillonite layers so that bending decreased (136). Iso-
morphous substitutions (chapter I) within the clay layers
are partly responsible for the extensive clay undulations,
as substitution of a larger cation for a smaller one will
alter the mismatch between octahedral and tetrahedral
layers (15). The intercalation of a charged pillared spe-
cies in the galleries, then modifies the interlayer charge
distribution so that oxygen framework.distortions decrease
and the layers become cemented together.

Figure 17 shows a specimen of Fe-pillared‘Wyoming
montmorillonite that was prepared in the same manner as in
Figures 16a and b. The increase in ordering of clay layers
is clearly evident in these micrographs. Undulations have
been greatly minimized through pillaring, though some
bending and curling near edges are still present, as indi-
cated in Figures 17b and c.

When the focused electron beam in the microscope

impinges on a specimen surface, a wide variety of

80

Figure 16. Edge—on SEM view (a) of Wyoming Na+-
montmorillonite air dried on glass. SEM image
(b) near a crack in the air dried clay.

81

 

82

Figure 17. SEM images of Fe/PILC air dried on glass
at various magnifications. The substantial decrease
in layer bending is evident at all magnifications.

83

 

84

interactions occur, including the emission of secondary
electrons, backscattered electrons, characteristic and
continuum x-rays, Auger electrons, and other assorted pho-
tons (137). These signals originate from specific volumes
or depths within the sample that are a function of average
sample atomic density, and the initial electron beam
energy. The specific emission volumes are illustrated in
Figure 18. Characteristic x-rays can be produced when the
incident beam dislodges an atomic inner shell electron,
resulting in an atom in a excited state. When the atom
returns to the ground state by transition of an outer shell
electron to fill the inner shell vacancy, the atom loses
energy by the emission of an x-ray photon. Due to quant-
ized atomic energy levels the emitted x-ray will be charac-
teristic of each excited atom. Characteristic x-ray emis-
sion offers the capability to perform 22.2122 elemental
analysis with microscopic resolution to identify elements
and investigate their distribution.

Figures 19a and 20a show electron micrographs of air-
dried iron-pillared clay samples. .A computer controlled
scan of the electron beam accross the sample, and analysis
of the characteristic x-rays produced, will give the dis-
tribution of counts recorded for each element. Figures 19b
and 20b show these linescans between points indicated on
the neighboring micrographs. The linescan in 20b gives a
qualitative indication that regions of high iron

concentration corresponded with regions of low Si and Al.

85

Electron
Beam

10—15 3 Auger electrons

Secondary electrons

Backseattered electrons

 

 

 

Characteristic x-rays

Figure 18. Generation of various signals in samples due to
interaction with the electon beam in electron microscopes.

86

Figure 19. SEM image of Fe/PILC (a) and the elemental
distribution (b) determined by an x-ray microanalysis
linescan.

Figure 20. SEM of a different area (a) and an accom-
panying linescan (b) between the arrows.

87

 

 

 

 

 

 

 

 

88

A linescan including the region near an edge (Figure 20b)
shows a high concentration of iron near the edge relative
to Si and A1. The Si and A1 distributions are indicative
of relative clay concentrations. The fact that iron was
found in excess where clay was in less abundance was unex-
pected. This pointed to the fact that iron deposits may
exist within the clay and at the edges. The calcined
Fe/PILC samples were kept under room temperature conditions
for the months between synthesis and SEM investigation.
That inhomogeneities in Fe distribution were evident was
interesting as the Fe/PILC used exhibited relatively well—
defined XRD patterns after calcination. We postulate that
redistribution of Fe occurs in these clays, and this may
have particular importance in explaining the steady
increase in a with time in the catalytic studies. The
implications and validity of these qualitative inferences
were further investigated.

When x-rays are induced by electron beam.bombardment
they are partially absorbed as they travel from their point
of emission to the surface. Based on the 20kv accelerating
voltage and the density of clay the maximum depth of x-ray
production is near 6-7 microns in these studies. x-ray
production in SEM is a function of the sample topography,
as the electron trajectories and path length are dependent
on the angle between the incident beam and sample. The
elemental distributions shown in the linescans then are

qualitative as the corrections for x-ray absorption, atomic

89

number and fluorescence effects required to quantitate
results (138) were not applied.

To obtain a better picture of the validity of the
linescan interpretations, the Na+montmorillonite shown ear-
lier, and a sample of thermally, stable, chromiumrpillared
clay prepared by S.D. Landau (104) were studied. Figure 21
shows an SEM micrograph of Na-montmorillonite along with a
linescan between the points indicated. The linescan ex-
hibits severe fluctuations in Si and Al x-ray count rates
as the beam traverses the peaks and valleys of the sample.
X-rays are nearly completely absorbed in regions of severe
depression, yet the relative concentrations of Al and Si
correspond with one another. Iron here is near the thres-
hold level of detection as the only iron in natural mont-
morillonite is that found due to substitutions in the octa-
hedral layer (about 4 wt% reported as Fe203). In Figure 22
a linescan between the points indicated in Figure 16a is
presented showing less severe x-ray count fluctuations.
Although the x-ray counts vary with morphology, they do so
in a consistent manner. It should be reinterated that such
fluctuations should be far less with Fe/PILC than with
Na+montmorillonite as the dried film undulations are far
less severe.

Figure 23 illustrates the linescan found on a basal
surface of a highly ordered chromium pillared montmoril-
lonite, which supports the argument that pillared clays

should show a homogeneous distribution of elements in

90

Figure 21. SEM image of the undulating surface of Na+
montmorillonite (a) and an x-ray microanalysis line-
scan (b).

Figure 22. Elemental microanalysis of Na+-montmori11-
onite between the arrows in Figure 16a.

Figure 23. Elemental linescan of a well ordered Cr-
pillared montmorillonite. Here the Cr distribution
follows that of Si, indicating a homogeneous pillared
clay.

 

 

 

0“.

21b

 

 

"...! '

 

91

Fe

 

 

 

 

 

 

 

 

 

 

92

qualitative x-ray microanalysis.

E. x-Ray Diffraction

At this point it was evident that heterogeneities in
Fe distribution in Fe-pillared clays did indeed exist and
that alteration of iron pillar structure and Fe distrib-
ution within the clay galleries was occurring with extended
aging under ambient conditions.

When the Fe/PILC used in the catalytic runs was ex-
amined by XRD, shown in Figure 24, it revealed no discern-
able reflections, suggesting that an amorphous material had
formed. Reexamination of the same XRD slide prepared after
Fe/PILC synthesis, which had shown a ordered, highly ex-
panded material after calcination, revealed that it had
become x-ray amorphous after 3 months under atmospheric
conditions. Apparently the humidity present in air causes
the iron pillar to hydrolyze, which results in an amorphous
material. Some of this iron then may agglomerate in the
reactor or SEM beam. This would explain the presence of
large Fe crystallites within the clay and iron being trans-
ported to the edges of individual clay platelets. The
increase in a then is most likely a result of the small Fe
active sites on the pillars decaying with time and large Fe
agglomerates being formed.

Aluminum and zirconium pillared clays are typically
stable after calcination with respect to this reversal of

the pillaring process but many other pillars exhibit pillar

93

Calcined

350°C Air

 

After Exn. f

Initial Calcined Fe/PILC ‘
After 3 mo., Air

1 1 I I j j
12 10 8 6 4 2
Oligrcees 213

 

Figure 24. XRD patterns of Fe/PILC before and
after Fischer-Tropsch reaction. The initially
well ordered material (top) was essentially
XRD amorphous after 3 months in air.

94

destruction at elevated temperatures. Indeed, pillar
hydrolysis even limits the application of Al and Zr pil-
lared clays for catalytic cracking due to steam induced
pillar hydrolysis.

The fact that iron pillared clays show an instability
towards pillar hydrolysis and layer collapse at the rela-
tive humidity of air strongly suggests that synthesis of
Fe/PILC stable under the conditions of FT, where water is
produced, would be improbable. Thus, more stable pillared
clays such as Al-pillared clays were investigated as means
to supporting catalytically active FT metals. Even though
the iron pillar did not have the long term stability ini-
tially hoped for, the pillar was certainly active in FT
initially. The novel idea of confining the active metal
between clay sheets in the form of a pillar demonstrates
that pillars themselves may be considered as catalytically

active components.

F. Conclusions.

The iron pillared clays used here offered unique cata-
lytic sites for Fischer-Tropsch (FT) hydrocarbon production
as the metal particles were very small and confined within
the microporous cavities of the pillared clay. These iron-
pillared clays had higher basal spacings at elevated tem-
peratures than any previously or presently reported Fe in-
tercalated clay system. The high gallery heights of these
pillared smectites (15 A), interlayer surface areas, and

initial thermal stabilities suggested that these materials

95

had considerable potential for investigating unique cataly-
tic properties of the small iron active sites. The unusual
catalytic site in these systems was the pillar itself.

The pillar was indeed active in FT converting carbon
monoxide to C1-C7 hydrocarbons. Straight chain paraffins
and olefins were the dominant hydrocarbon products,
although methanol, ethanol and propanol were also detected.
Water was the major component of the trapped liquids (87
wt%) with the remainder composed of the light alcohols.

Product distributions primarily followed the Schulz-
Flory (SF) relationship with significant deviations in the
C1-C4 range being present at short run times. These devia-
tions were common however of the FT process, such as high
methane and low C2 production. Deviations from SF
decreased significantly with longer run times, and by 20
hrs, hydrocarbon production showed no deviations. Olefin/
paraffin ratios increased with greater CO conversion in
agreement with previous supported iron catalysts.

The probability of chain growth.(a) derived from SF
plots of hydrocarbon distributions slowly increased with
time signifying the trend toward higher hydrocarbon produc-
tion. The decrease in a with time was aided by the drop
in methane production from 50 wt% at 60 min, to near 37 wt%
at greater than 100 min of reaction. This propensity for
higher hydrocarbon production may have been related to
structural or compositional changes of the active site with

time. This possibility was supported by reports of two

96

chain growth probabilities on supported Fe-catalysts, that
may have active sites differing in structure or relative
degrees of promotion by potassium.

Characteristic x-ray production induced by the elec-
tron beam in a scanning electron microscope was used to
examine the distribution of elements within these pillared
clays. This microanalysis indicated heterogeneities in the
iron distribution. Regions of high iron concentration
corresponded to low concentrations of the Al and Si of the
clay. Additionally concentrations of iron were enriched at
edges relative to basal planes which suggested iron migra-
tion. Natural montmorillonite and.a*well-ordered.chromium
pillared clay were used as standards to test the validity
of the qualitative microanalysis. 'The implications of the
fluctuation in Fe distributions of Fe/PILC were supported
by this work.

This led to reexamination of the iron pillared clay
used in catalysis and the material that had not been sub-
jected to the harsh reaction conditions. X-ray diffraction
patterns of these materials revealed that iron-pillared
clay had collapsed in the reactor, when stored at atmos-
pheric conditions. As the relative humidity present in air
induced pillar hydrolysis and reorganization of the
Fe/PILC, synthesis of iron-pillared clays that would be
stable under FT conditions (where H20 is produced) was
deemed improbable.

Although catalytically unique, the Fe pillar

97

demonstrated activity in FT hydrocarbon production, the
demise of the pillared structure with humidity led us to
investigate other means of highly dispersing active metals

in the microporous regions of pillared clays.

CHAPTER IV

Ruthenium on Alumina Pillared Montmorillonite

Clays pillared with aluminum oxycations have generated
more interest and research since their discovery than any
of the other inorganic pillars mentioned in Chapter I.

This is due to their excellent thermal stability, Bronsted
and Lewis acid sites, and adjustable pore structure.
Pinnavaia and coworkers (139) demonstrated that the method
used to dry the flocculated pillared clays was more impor-
tant than whether the Al-oligomer was obtained from base
hydrolyzed Al solutions or a commercial reagent (aluminum
chlorhydrate). In this study air drying of pillared clay
solutions yielded materials with zeolite-like micropores
whereas freeze dried clays had considerable macropore char-
acter. They suggested that the presence of macroporosity
in freeze dried clays was due to delaminated regions, where
edge to face and edge to edge association of clay layers
occurs, based on similarities to a synthetic lath shaped
hectorite (Laponite) that exhibits strong delamination
tendencies. The presence of delaminated regions may alter
the diffusion properties of reactant molecules, allowing
larger molecules to penetrate the clay interior for cata-

lytic reactions. The importance of this aspect of pillared

98

99

clays has been further investigated by catalytic (140,141)
and physical characterization (142) studies.

Although the delamination model explains numerous dif—
ferences between air dried and freeze dried clays, recent
work by Van Damme and Fripiat (143) suggests another expla-
nation. They reanalyzed earlier adsorption data for freeze
dried and air dried clays by a fractal model and found that
pillars are homogeneously and regularly distributed in both
of these alumina pillared clays, and not concentrated in a
zeolite-like region relative to a delaminated area. An
alternative explanation for these observed differences is
that continued pillar hydrolysis occurs in air dried clays
filling additional porespace, whereas in freeze dried clays
hydrolysis is halted due to the temperatures employed and
the rapid withdrawal of water (144). This is supported by
the fact that when Al-pillared clay solutions were washed
only 2-4 times and then dialyzed, adsorptions of perfluoro-
tributylamine and surface areas were lower than pillared
clays washed and centrifuged numerous times (104). Both
clays were air dried and had similar gallery heights. 'The
size and distribution of pores within pillared clays has
particular relevance to catalytic studies with these mate-
rials, as the dimensions and populations of constrained
cavities are affected.

The interaction between the pillars and neighboring
clay sheets has also been an area of active research in the

past few years. Plee gt al.(145) showed that a

10()

tetrahedrally charged smectite (beidellite) interacted with
the clay layers when calcined, whereas octahedrally charged
smectites failed to react when treated similarly. The au-
thors suggested that the substantial structural transforma-
tion for beidellite arose from the formation of a 3 dimen-
sional network as the pillar was grafted onto the clay
structure. The resulting materials had high surface areas
and increased acidity relative to octahedrally charged
clays. 'The crosslinking of pillars to the clay network may
result in enhanced thermal stabilities and increased resis-
tance to pillar hydrolysis which would also be of interest
in catalysis.

The use of well defined molecular clusters in loading
metals on catalyst supports has gained considerable accep-
tance in the past decade due to the advantages mentioned in
chapter I. The discovery by Giannelis (103) that metal
carbonyl clusters can react with the intracrystalline
acidity of alumina pillared clays is of particular interest
as it allows metals to be highly dispersed in clay gal-
leries. For the present study the fact that alumina pil-
lared montmorillonite (APM) was found to protonate
Ru3(CO)12 was especially interesting as ruthenium is one of
the most active FT metals and has a propensity for forming
high molecular weight hydrocarbons.

Giannelis found that Ru3(CO)12 reacted with the acid
sites of activated APM, forming (HRu3(CO)12+/APM), which

was identified by infrared spectroscopy. IRemoval of the

101

protonated cluster from the interlayer by exchange with
KPF6 in acetone, allowed the cluster to be further identi-
fied and confirmed by agreement with literature results.
Experiments with collapsed clays exposed to Al-polycations
showed that Ru3(CO)12 clusters were only physisorbed and
could be removed by CHZClZ washings. The protonation of
this carbonyl cluster in homogeneous solution requires 98%
sulfuric acid, which again demonstrates the intracrystal-
line acidity of APM. Cluster protonation within the gallery
however, was reversible and after a few hours in air the
infrared absorptions were identical to Ru3(CO)12. Pro-
longed exposure to air, or vacuum treatment at 25°C, led to
mononuclear ruthenium complexes of the type [Ru(CO)2(O-Al<
)ZJn' When heated in air, further decarbonylation occurred
and oxidized Ru atoms were formed.on the clay mineral.

The objective of this present study was to charac-
terize the catalytic features of these ruthenium loaded
clays in FT. The dispersal of Ru3(CO)12 in APM due to the
intracrystalline acidity was particularly important as the
ruthenium was specifically contained in micropores where a
constrained environment would exist for the FT site. Thus,
the distribution of hydrocarbons obtained in FT catalysis
was closely examined for potential influence of the con-
strained site or chemical nature of APM.

The breakup of the trinuclear Ru carbonyl skeleton and
the eventual decarbonylation to Ru atoms suggests that Ru

mobility on the clay surface would have to be examined as

102

metal sintering may occur. The most effective means of
characterizing the change in Ru disposition was high
resolution electron microscopy. A brief introduction is
given here, although the reader may wish to examine the
references given for additional information.

High resolution electron microscopy is one of many
high vacuum techniques that have been employed extensively
recently for the characterization of catalyst surfaces.
The imaging and microanalysis capabilities of this tech-
nique have provided important information concerning the
composition, structure, and disposition of the fundamental
particles that constitute heterogeneous catalysts (146-
148). Recent efforts in this field have focused in part on
the imaging and analysis of metal crystallites of various
supports (149).

Although scanning electron microscopy (SEM) and trans-
mission electron microscopy (TEM) can provide a great deal
of information, the most powerful and versatile EM tech-
nique for catalyst characterization is analytical electron
microscopy (ABM). The dedicated AEM used in this work (a
Vacuum Generators HB501) has a very bright (yet small)
electron beam, ultrahigh vacuum specimen area (10'9 to 10'
10 torr), and a variety of imaging and analysis capabili-
ties. The specimen chamber and arrangement of various
detectors in the VG H8501 is illustrated in Figure 25. In
addition to the analysis methods shown, microdiffraction

patterns can also be recorded, and the availability of

103

fiCANNING TRANSMISSIM ELECTRON 11100050091:

 

 

 

 

 

 

 

| tags:
mom-nan memoirs:
1m: macros r -
UNDEVIATED
*,,r”asmms
max-mm
we mama
sunmmuw
mwummasmus

 

Rum

14M ‘\\\\\
SPECMTER m0 (LEW

 

 

 

 

 

 

 

 

 

 

 

IEAI
(LICTIOI IEAI
SCAlllED Ill A
“STE! OI mu
1°10,
Edam
hen
L .f J
Elem —-0

 

 

 

Figure 25. Arrangement of detectors in the specimen
chamber of a dedicated STEM. Reproduced from ref. 148.

Table 7. Summary of Physical Data for Al-Pillared Clays.

 

 

 

° Surface PFTBA
_XBDVdeoi(A17 Area Adsor tion mmol’ ‘
This work 18.4 345 0.41
Landau (104) 20.1 369 0.42
Tzou (58) 18.5 260 0.24

 

clays calcined at 350°C
PFTBA = perfluorotributylamine

104

electronic processing allows the signals from the dark and
bright field detectors to be combined.

Dark field images are of particular importance in EM
investigations of supported metals as the electrons de-
flected at high angles by supported metals are preferen-
tially recorded in dark field.

A. Physical Properties of APM and Ru/APM

The APM obtained from the synthetic method described
in Chapter II had the physical properties listed in Table
7. These properties were similar to those previously
obtained by Landau (104) and exhibited higher surface areas
and PFTBA adsorptions than comparable materials prepared by
Tzou (58).

Elemental analysis showed that ruthenium loading
ranged from 0.2 to 0.5 wt%. Characterization of these
materials by infrared spectroscopy (Figure 26a) revealed
that they were protonated Ru clusters bound within the clay
interlayer, in agreement with the previous work mentioned
(103). When the Ru/APM samples were weighed and loaded
into the reactor tube they were still green. Infared
spectroscopy of these materials prior to catalysis revealed
that the protonated complex had undergone decarbonylation
in the dessicator (Figure 26b) to form a complex similar to
that previously assigned to monomeric [Ru(CO)3X2]n/A1203 by
Kuznetsov and coworkers (150). Although the APM was not
examined _i_._r_1 gttt by infrared in the present work, a few

post-catalysis samples were characterized by infrared as

105

2200 2000
L___J

Figure 26. Infrared spectra of Ru/APM (a)
immediately after synthesis (2130, 2104,
2080 cm‘l), (b) prior to catalysis, aged

1 month in dessicator, (2150, 2080 cm’l),
(c) post catalysis after 12.5 hrs., 225°C
rxn. (2100, 2056 cm'l). a. ordinate expan-
sion 400x.

106

shown in Figure 26c. These absorbtions are at higher wave
numbers (2100, 2056 cm'l) than those assigned by the above
mentioned authors to [Ru(CO)2X2]n/A1203 (2045-2050, 1950-
1970 cm'l) and may be due to CO adsorption on Ru crystal-
lites. When the post-catalysis Ru/APM samples were examined
by XRD the reflections observed indicated that APM was
stable under FT conditions, in contrast to Fe/PILC. For
example, a typical APM used in this study had a d001 -—-
18.4A. After this material was loaded with the Ru cluster,
prereduced in the reactor at 420°C for 5 hrs., and run 30
hrs. at 275°C in FT synthesis, it still gave an 17.8 A d001
reflection. Thus despite these harsh conditions, and the
production of H20 in FT alumina pillared clays were stable

and retained their expanded gallery heights.

B. Pristine vs. Prereduced Ru/APM Catalysts

As mentioned earlier, metal carbonyls provide a low
metal oxidation state when used as catalyst precursors.
Although it is highly probable that support interactions
partially oxidize the Ru in these clusters, completely re-
duced Ru decomposition products have been claimed (151) for
Ru3(CO)12 vapor impregnated zeolites. The dispersion of
these metal complexes in zero-valent, or low oxidation
states that can be readily reduced, offers the advantage
that catalyst prereduction can be avoided. This is desira-
ble as the elevated temperatures employed to reduce metal

catalysts may induce metal migration and agglomeration.

107

To examine whether the freshly prepared Ru/APM cata-
lysts would be active in FT (indicating low valent Ru ini-
tially), a number of catalysts were examined without
prereduction in hydrogen. Although C1 to C5 hydrocarbons
were detected for catalysts without prereduction, conver-
sion of CO was often very low (0.4%) at one atmosphere and
275 °C. Elevated pressures resulted in increased conver-
sion. At short run times (8-45 ndJn) unusual FT product
distributions were obtained that showed significant devia-—
tions from common Schulz-Flory distributions. .Among the
uncommon features observed were low methane production
(26 wt%), relatively high yields of (°6'C10) hydrocarbons,
and the appearance of more than one chain growth probabili-
ty. These features were short-lived however, and after
approximately 2 hours: no deviations from SF were found,
methane yields increased dramatically, and production of
higher hydrocarbons decreased. Although conversion in-
creased substantially with longer reaction times, this was
normally due to the increase in methane production.

Figure 27 shows a particularly interesting example of
a Ru/APM Fischer-Tropsch catalyst that was ggt prereduced
in hydrogen prior to catalysis. 'This experiment showed an
SF distribution of C1-C5 hydrocarbons and featured unusu-
ally high production of C6‘C10 hydrocarbons. These devi-
ations were slightly diminished at a reaction time of 43
min however, and by 138 min the product distribution fol-
lowed that predicted by SF with only two exceptions. At

108

 

d

C

C)
a A mean;

A

:1
.9

[>0
0 D/
e/

 

 

 

001; \A
= I
>- 1
3 ‘ (3 (A
A
0.001.
3 \
1 A
‘ O
O
. C><D C) (D
1 35 1'0

Carbon number (n)

Figure 27. Schulz-Flory plots of Ru/APM
catalysts after 8 min. 0 , and 138 min. A of
Fischer-Tropsch. The catalyst was ggt pre-
reduced prior to reaction. Reaction condi—
tions: 275°C, 8 atm., 300 hr'l, 82/60 = 2.

109

even longer times on stream the SF plots obtained were
similar to that at 138 min. Conversion for this reaction
at 8 atm was19.8% so that production of the unusual quanti-
ties of C6'C10 hydrocarbons was substantial and easily
detected by gas chromatography. In this case methane se-
lectivity decreased with time.

The most interesting feature of the deviations in Fig-
ure 27 is that the product distribution appears to follow a
very different chain growth probability for C1-C5 as com-
pared to C6-C10 hydrocarbons. Although the presence of two
chain growth probabilities in FT has been reported for
iron-based catalysts, as mentioned in Chapter 3, ruthenium
catalysts have not exhibited this phenomenon. The initial
heterogeneity of the Ru active site may result in a variety
of geometrical structures of CO attachment which result in
different activation energies for reaction intermediates,
and enthalpies of absorption for reactants and products
(152). A number of studies (153-155) indicate that CO can
adsorb on ruthenium in monoadsorbed and diadsorbed.(CO-Ru-
CO) forms. The diadsorbed form was associated with iso-
lated Ru atoms, or very small clusters which interact
strongly with oxygen from the support, in earlier work
(150,155). The geometry of CO adsorption may have partic-
ular importance as experiments by Kellner and Bell (155)
suggested that sites that monoadsorb C0 are active in FT
whereas diadsorbed CO sites were inactive.

In experiments without a prereduction step, H2 and CO

110

a: «so w
00 _OI sm sass
amass asmussaa\vososssa _ _
Guess saunas:5\s_uo_c A v
mOh cal on as case __<

 

 

 

 

_N.._ _N.—_ _oo.o_
2:.e. .e.n. «s.s. .o.e.
... 8.. e.e c.”— «.e. s.s. s.s. v.6 e.e has ee;==:.=te
_vb._ _v._. —pc.e_
.s.ec Ae.~. 2o.~c
... 3.: o.v N.v —.a ~.~— o.o> o.N o_. new couscoscsa
_m.e_ _ve.e_
Ac... An.o. 2....
v._ N." v.v h.v~ (.53 9.0 h.N new voosvosscc
.46 so .o .o no «c .o Auvaeaeeopeeo use: Awe—v Aooc.esoe sea-easteet;
* .Jo.s; flowaoooz

 

.10. s «a cane uses—asap asexrn .o oases

111

flows were equilibrated over the catalyst at room tempera-
ture and 120 psi before the temperature was increased to
275°C. It is likely that decomposition of the carbonyl
cluster in H2 and CO gave Ru active sites dissimilar to
those found under steady state conditions. The changes in
product distribution, chain growth probability, presence of
SF deviations and conversion are probably a result of the
active site reaching a homogeneous, steady state structure
with time.

A comparison between catalysts that were not pre-
reduced with those that were reduced at 1 atm revealed
significant differences as demonstrated in Table 8. Ru/APM
catalysts prereduced in H2 at 420°C gave greater conver-
sions, reaction rates, and higher hydrocarbons than the
pristine catalysts for reaction conditions of 1 atm. and
275 °C. Small amounts of C5 hydrocarbons were obtained
with nonprereduced catalysts whereas prereduced.Ru/APM gave
significant yields of hydrocarbons up to C8. For pre-
reduced catalysts methane production was lower than that of
nonprereduced catalysts. .Although only a limited number of
data points were obtainable for olefin/paraffin and
branched straight chain ratios, these values are given in
Table 8 to illustrate that olefin/ paraffin ratios are high
and branched hydrocarbon production is minimal for light
hydrocarbons. Representative data for prereduced Ru/APM
catalysts reacted at lower temperatures is also included

here as it shows that conversion decreased substantially.

112

The Schulz-Flory plots shown in Figure 28 demonstrate
that higher hydrocarbon production is favored for the pre-
reduced catalysts. The yields of the various hydrocarbons
produced fall on a straight line for all three run condi-
tions, obeying SF. Further, the slopes for both prereduced
runs are clearly similar, but these slopes differ from the
catalyst that was not prereduced. Additionally, it should
be pointed out that methane production decreased substan-
tially when.Ru/APM catalysts were run at 225°C as compared
to higher reaction temperatures. IMethanation is favored at
higher temperatures, as shown in the simple scheme of run
variables in Chapter I, so this result is expected.

What was not expected is the relatively large differ-
ence between the arvalues for nonprereduced catalysts com-
pared to their prereduced counterparts. The relationship
between these plots does not change significantly in the
first few hours on stream either, suggesting that a signif-
icant difference between the steady state nature of nonpre-
reduced and prereduced catalysts. Additionally, the devia-
tions from SF distributions found initially for nonprere-
duced catalysts were not observed for prereduced catalysts.

Infrared investigations of Ru3(CO)12 decomposition
products on A1203 as a function of treatment conditions
(150, 156-158) have shown that surface aluminates and
microcrystallites of Ru form in H2 at temperatures above
350°C. Kuznetsov 23.21; (150) investigated ruthenium

carbonyl clusters of varying nuclearity and found that all

Wn/n

113

 

A AAmLA

[36>

- . .....9
D

0.011

A AJA‘A

 

0.001

4 A A LAAAA

 

 

 

T

T i :3 4 5

m4

Carbon number (n)

Figure 28. Comparison of Schulz-Flory plots

for prereduced O and nonprereduced A catalysts
at 275°C. A prereduced catalyst run at 225°C
gave the distribution shown by E] . Reaction
conditions: 1 atm., 300 hr’l, 50 min. T08.

114

metal precursors decompose at room temperature in air (or
vacuum elevated temp.) to a form that can be represented as
[Ru(CO)2(O-Al<)2]n. This is also the species postulated by
Giannelis (103) for Ru/APM at temperatures above 200°C in
vacuum. It would seem reasonable then that the Ru micro-
crystallites formed by prereduction in hydrogen at 275°C
and those formed without prereduction for Ru/APM at 275°C
in H2 and CO should be similar. Yet the fact that FT
catalysis of prereduced and nonpreduced Ru/APM is signifi-
cantly different suggests that the decomposition products
of [HRu3(CO)12]+ initially found within the micropores of
APM are substantially different for these two catalyst
treatments.

To investigate potential differences between pre-
reduced and nonprereduced catalysts high resolution ana-

lytical electron microscopy was applied.

C. Electron Microscopy

In order for valid conclusions to be made regarding Ru
crystallites observed by electron microscopy, potential
sample artifacts induced by the electron beam had to be
identified and eliminated. For this purpose freshly pre-
pared Ru/APM samples that had not been run under catalytic
conditions were examined. These studies showed that ini-
tially Ru microcrystallites were not clearly discernable on
the APM support at magnifications up to 1 million times, in
agreement with arguments stating that the microcrystallites

formed from carbonyl precursor decomposition are less than

115

2 nm (150, 156-158). IHowever, with time aggregation did
occur in the electron beam and large Ru crystallites were
found on the clay surface. Use of a liquid nitrogen cooled
sample stage in the microscope though, very effectively
minimized Ru aggregation as confirmed by additional studies
using fresh Ru/APM samples. These studies are described in
additional detail in Appendix A. IMicrographs shown in
throughout the rest of this text are taken from studies
using the liquid nitrogen stage. Initial microscope
studies of Ru/APM also revealed catalyst contamination by
the feed gas (caused by iron and nickel carbonyl deposi-
tion) and confirmed the successful elimination of this
contamination by appropriate purification procedures. This
aspect of the research is presented in Appendix B.

Representative STEM micrographs of nonprereduced cata—
lysts (159) are shown in Figures 29-31. The large Ru crys-
tallites present in micrographs are clearly evident as they
exhibit markedly different form and contrast as compared to
the diffuse image of the APM support. They are particu-
larly distinguishable in dark field images.

The Ru crystallites observed in numerous micrographs
of spent Ru/APM catalysts that were ggt prereduced occurred
over a broad size distribution as shown in Figure 32. For-
mulas for the statistical averages of Ru particle sizes
observed in numerous micrographs of these catalysts are

given below.

116

Figure 29. Dark field image of Ru/APM catalyst
after 30 hrs. of catalysis at 275°C showing Ru
aggregates. The catalyst was ggt reduced prior
to catalysis. Arrows indicate aggregate examined
by the EDS analysis in Figure 33.

Figure 30. Dark (a) and bright field (b) images of the
Ru/APM catalyst showing a coalesced Ru particle.
Catalyst was not prereduced.

Figure 31. Twinned Ru aggregate on a Ru/APM post-
run catalyst sample not reduced prior to catalysis.
The catalyst was run for 30 hrs. of FT catalysis.

117

 

118

RM APM ggt produced

 

 

 

 

 

 

 

 

 

 

161
dr‘2135.5run
12 4
ds :559.611m
3.
a-
4 .

 

 

 

 

 

 

 

 

 

 

 

 

 

10 3C) 5() 70 90 110
dilnnfl

Figure 32. Ruthenium crystallite sizes observed by
STEM for nonprereduced catalysts after 30 hrs. of
FT reaction at 275°C. Here n: is the number of
particles with diameter d:, dn is the number
average and d. is the surface average particle size.

119

dn = m (13 = Enidia

2 Di Enidiz

Here dn is the number average, dB is the surface average,
and ni represents the number of metal particles of diameter
di (nm). The surface average size for nonprereduced cata-
lysts is 59.6 nm. Based on the size, and contrast with the
clay support, it is clear that these Ru particles are
definitely located on the exterior surfaces of the pillared
clay. This distribution shows that large crystallites were
favored with nonprereduced catalysts as Ru.particles less
than 10 nm were not found. Metal particle coalescence is
apparent in Figure 30 and multiple twinning of small crys-
tallites is evident in Figure 31. These features are remi-
niscent of the crystallite migration model (160, 161) of
metal sintering where large crystallites move across the
support surface, collide, and merge into a single unit.

The complicated geometries of multiply twinned particles
result in surface sites that are unusual in catalyst parti-
cles, and may have particular importance in catalytic re-
actions that are dependent on surface crystallography
(162,163).

EDS microanalyses (Chapter II) were used in STEM
investigations to confirm the identity of small microcrys-
tallites, as the appearance of supported metals is a sensi-
tive function of numerous microscope parameters. The rep-

resentative EDS spectrum given in Figure 33 is from the

120

100

Eve-em

COUNTS

g3»

 

 

 

 

I
U
C
{L 1 . z
9 ML , 'p

Y Y " 7" Y Y r V I 1 Y Y Y Y ' V
5 0 10.0 15.0 20.0
IMIEGY (ks?)

0.0

Figure 33. EDS spectrum from the Ru crystallite
indicated in Figure 29. The Cu peaks are artifacts

originating from the Cu microscope grids.

 

 

 

 

 

 

56" m
Flu/APM prereduced
37‘” dn:5.6nn1
ds:6.9nn1
911
ll
‘45

dilnnfl

Figure 34. Ru crystallite sizes determined by STEM
studies on numerous Ru/APM particles after 30 hrs.
of reaction.

121

area indicated in Figure 29. It shows that this aggregate
on the clay support consists of pure ruthenium. EDS showed
that the occasional presence of Fe in spectra was due to
the presence of iron in the clay structure, which is typi-
cally 0.01 wt% based on bulk elemental analyses. This iron
did not migrate to the Ru aggregates as the iron intensity
was always identical to spectra taken from neighboring
regions without Ru aggregates.

STEM investigations, revealed however that the mean
size of the Ru microcrystallites for prereduced catalysts
was near 7 nm (Figure 32) nearly an order of magnitude
smaller than catalysts not prereduced. Additionally, the
distribution of particle sizes was nearly an order of
magnitude narrower. Figure 35 illustrates these crystal-
lites for a thin area of clay support. EDS analysis again
confirms their identity and purity as illustrated in Figure
35c. The annular dark field image clearly distinguishes
these microcrystallites from the support in thin areas but
the contrast in thicker areas is greatly diminished due to
increased scattering from the clay layers. The 10 nm
crystallite at the tip of the clay support in Figure 35
clearly is on the surface or edge of the clay layers,
however the number of Ru particles on the external surface
is relatively small. Surely if microcrystallites as small
as those imaged in Figure 35 were present on the surface
they would have aggregated during FT catalysis. Most of

the Ru is contained within the clay particles, as opposed

122

Figure 35. Representative ptgreduced Ru/APM catalyst
particle, found by STEM bright field (a), and dark

field images (b), showing small Ru microcrystallites. A
EDS microanalysis is given in Figure 35c. The catalyst
was run for 30 hrs. of FT at 275°C and 1 atm.

 

 

O I”)

 

124

6001

5

X

COUNTS

 

 

 

 

 

00 V - - 5p V V lmov 190' 2mo
ENERGY (Rel!)

Figure 35c. EDS spectrum from the 5 nm microcrystallite
indicated in Figure 35.

125

to the exterior dispersion observed for the catalyst which
was not prereduced (Figures 29-31).

The observation that prereduction results in more
efficient Ru dispersion, and that the majority of Ru crys-
tallites appear to remain within the microporous APM sup-
port is significant. Clearly there is a marked difference
in the Ru decomposition products obtained with prereduction
versus nonprereduction. The relative degree of Ru migra-
tion and agglomeration as well as location is quite
different with these two catalysts and may account for
catalytic differences already discussed.

The sintering of supported metals is commmonly
encountered in heterogeneous catalysis, thus the mech-
anisms, kinetics and measurement techniques of metal sin-
tering have received considerable attention. sintering,
and the design of catalytic reactors to minimize it, has
been reviewed recently by Lee and Ruckenstein.(164). Two
mechanisms account for the majority of phenomena observed,
and the mechanisms can occur simultaneously or singularly.
According to the crystallite theory (161), individual metal
crystallites move across the support surface, and merge
into larger crystallites when they come close. The
features shown in Figure 30 of large Ru crystallites in the
process of coalescing on the surface of APM, clearly shows
that this process is occurring. The atom migration model
(160) conversely explains that crystallite growth is caused

by the diffusion of single atoms or small molecules on a

126

substrate. The formation of the small Ru microcrystallites
on prereduced catalysts, and the migration of Ru from APM
micropores, suggests that this mechanism is also operative
in Ru/APM catalysts. sintering is promoted by elevated
reaction temperatures due to enhanced thermal motion of
atoms, which makes the energy liberated by exothermic reac-
tions (i.e. FT) especially important. Conversion was norm-
ally kept below 7% in this study to minimize such tempera-
ture affects.

Atom migration and sintering during reaction affects
the metal dispersion, geometrical structure of the sup-
ported metal, and degree of electronic interaction with
support. The conversion increase observed for nonpre-
reduced catalysts is most likely due to the relatively
rapid migration of highly dispersed Ru atoms into micro-
crystallites and eventually the large crystallites observed
after 30 hrs. of reaction in the STEM studies. King (165)
found that the specific activities for methanation and CO
consumption increased.monotomically with decreasing metal
dispersion, which supports the conversion increase observed
with Ru/APM. Kellner and Bell (96) studied 1.3 wt%
Ru/A1203, prepared from Ru3(CO)12, in a glass microreactor
capable of measuring synthesis activity and H2 chemisorp-
tion isotherms immediately following catalysis. They de-
termined that as the dispersion fell from 0.9 to 0.6, due
to Ru particle growth, the specific activity for hydro-

carbon production (activity per exposed metal atom)

127

increased dramatically. The results of Ohukara gt gt;
(166) further support this inverse relationship between
specific activity and dispersion. They suggested, based on
infrared evidence, that this is due to the increased rela-
tive concentration of the tightly bound, inactive
diadsorbed CO mentioned earlier with smaller Ru particles.
Although Ru/APM catalysts were run for 30 hrs. with reac-
tion procedures different from these studies, for com-
parison the final dispersions of prereduced and nonprere-
duced catalysts would be 0.2 and 0.03, respectively
(dispersion - l/d(nm), from Figures 32 and 33, assuming
spherical particles).

The probability of chain growth was independent of
dispersion in the work of Okuhara gt git, but Kellner and
Bell found that it increased slightly with decreasing dis-
persion. The increase in <1 shown in Figure 28 for a non-
prereduced catalyst agrees with this latter relationship.
The increased selectivity for methanation with even longer
run times does not. The lack of chain propagation at ex-
tended periods may be the result of the large crystallites,
present in nonprereduced catalysts, oxidizing as they grow
forming active sites that can only produce methane.

The decrease in SF deviations with time is certainly
due to the relatively rapid migration of Ru for the nonpre-
reduced catalysts. Whether the active site of nonpre-
reduced catalysts was unusual in structure, or electronic

interaction with the support is overshadowed by the fact

128

that with nonprereduced catalysts Ru readily vacates the
micropores of the APM. The most important feature of the
Ru/APM catalysts is that the ruthenium is selectively dis-
persed in micropores where the Bronsted acid sites respon-
sible for the Ru3(CO)12 protonation are located. The loss
of this distinguishing feature for nonprereduced catalysts
makes them of diminished interest relative to the Ru micro-
crystallites retained within the micropores for prereduced
catalysts.

The molecular weight distributions and chain growth
probabilities afforded by prereduced Ru/APM catalysts are
very similar to the above-mentioned studies with highly
dispersed Ru on A1203. Very few deviations from SF were
observed for prereduced Ru/APM, which is also consistent
with the Ru/A1203 studies. These similarities suggest that
prereduced Ru/APM is in fact very similar to Ru/A1203 and
that no unusual deviations due to the steric confines of
the micropores are observed, even at short run times. As
mentioned in Chapter I, increased hydrocarbon chain length
has a much smaller effect on the diffusivity of paraffins
than that of branching. Chain limitation has been claimed
as a characteristic feature of active sites confined in
zeolite cages (Chapter I), yet these SF deviations usually
are short-lived, because metal migration alters the initial
active site. This is, of course, similar to what is ob-

served here.

A few other points should be mentioned here concerning

129

the nature of the pillared clay support. The two-dimen-
sional pores of pillared clays are less rigidly defined as
they are the result of pillars placed between two-dimen-
sional sheets, rather than the three-dimensional channel
networks of zeolites. These pores are also not as well-
ordered as their zeolite counterparts. When pillared clays
are air dried on glass the layers stack parallel to the
glass in a regular fashion for the most part. As they dry,
surface tension forces associated with the evaporating
water promote face to face interactions. ‘Yet, occasional
layer packing disorders introduce larger pores into the air
dried structure than expected based on the gallery height
from XRD.

We propose that the Ru microcrystallites, observed
by electron microscopy, for prereduced samples are formed
in interlayer voids created by layer packing disorders, and
within the original clay gallery micropores. In this
model, illustrated in Figure 36, the larger Ru particles in
the range 2-6 nm are accommodated.in the mesopores formed
by imperfect packing between adjacent clay layers and by
folding of the clay layers. On the other hand, the smaller
Ru aggregates (g 2 nm) are of appropriate size to be accom-
modated in the galleries defined by the pillared clay
layers. .Additional support for this model is found in the
pore size distribution shown in Figure 37, which was deter-
mined by nitrogen desorption isotherms at -196°C for the
APM used in catalyst studies. .Although the majority of the

130

 

 

 

 

 

 

 

 

 

  
    

o 0 O I D. CL .D l
o “lo cmlqzummhmmmw
0
0-0-0 ,0-0-0-0';o .-
OIlWllfldhllflllo A'Ih'hih'hll:ll:'
‘—
2 e o e s 0% _I-s-s-e sIs .-
s s e s o s omnmmo 222.9 n m

a 0.0-0.0.0

  

 

 

 

 

 

 

Figure 36. Model of air dried pillared clay showing
mesopores created by layer packing disorders and layer
bending. Plausible locations of Ru microcrystallites
>2nm (labeled Ru), and less than 2nm (filled triangles)
are indicated. Slabs represent clay layers and small
open circles the ”alumina" pillars.

7.0.10

AVp/A

42

136

(30

24

18

12

Figure 37.
nitrogen desorption isotherms.
pore radius,

 

 

 

131

 

 

Pore size distribution of APM from liquid
Here F} is the mean
and A5Vp is the change in pore volume.

 

60 80100

132

pore volume is due to micropores with diameters near 20A,
mesopores with diameters near 40 A are also present.
Apparently, reduction in hydrogen limits the migration
of the ruthenium to these two regions of the microporous
support. However, when the Ru/APM is not prereduced,
Ru3(CO)12 may be formed and migrate to external surfaces
where it decomposes and eventually Ru atoms aggregate into
very large crystallites up to several hundred nm in dia-
meter. Metal sintering on a support is dependent on the
magnitude of surface interactions and chemical environment,
among other factors. It is likely that the initial pres-
ence of CO in nonprereduced catalysts induces the formation
of metal carbonyls which have enhanced volatility. Con-
versely. prereduction in hydrogen at elevated temperatures
probably causes the formation of Ru microcrystallites
within the micro- and mesopores. 'These microcrystallites
may then be trapped within the support and not as mobile
when CO is introduced for FT reactions. Goodwin 22.21;
(167) found that a 3.8 wt% Ru/A1203 catalyst lost up to 40%
of the metal when exposed to 1 atm. CO under flow condi-
tions for 24 hrs and zoo-265°C. They identified Ru(CO)5 as
the volatile species at 25°C and Ru3(CO)12 at temperatures
above 200°C, by infrared characterization of a solvent
trap. Clearly such a reversal of the original metal
loading process would account for metal migration from
micropores, transportation through the catalyst bed, and

deposition and aggregation on clay surfaces.

133

D. Nature of Ru microcrystallites.

Direct evidence for microcrystallites within thicker
regions of the APM support would give evidence to the model
stated above. The characterization of the nature of these
microcrystallites would also provide useful information.

The identification of Ru microcrystallites > 5 nm is
readily accomplished by dark field imaging, but thickness
variations and electron scattering by the APM support di-
minish the effectiveness of dark field imaging for smaller
microcrystallites. In many cases conventional bright field
images lack sufficient contrast for positive identification
of small, supported metal particles (168). The z-contrast
imaging technique (169) however, seems well suited for the
detection of small metal crystallites and even atoms of
high atomic number on light weight supports. This tech-
nique of electronically combining the elastic and inelastic
signals recorded by the dark field detector and electron
spectrometer, respectively, changes the dominant contrast
mechanism forming the final image (169L. The ratio or
difference of these two signals supresses background varia-
tions due to variations in support thickness or electron
source intensity so that detection of heavy atoms is en-
hanced. Figure 38 shows the improved contrast between Ru
microcrystallites in the z-contrast ratio image as compared
with a normal dark field image. Aggregates less than 5 nm
are evident in the z-contrast image which are not discern-

able in the dark field image. iFigure 39 compares a bright

134

Figure 38. Dark field (a) image of Ru/APM after catalysis.
The area indicated in (a) is shown below in (b), using the

Z-contrast ratio imaging technique, which shows microcryst-
allites < 5nm. Catalyst prereduced.

Figure 39. Bright field (a) and difference Z-contrast image
(b) for a prereduced Ru/APM catalyst particle showing the
ability of Z-contrast to image Ru in thicker clay regions.

135

 

136

field image and the z-contrast difference image. Clearly,
z-contrast greatly simplifies the identification of the Ru
domains and demonstrates Ru microcrystallites are present
within thick sections of.APM. The inelastic signal from
the pillared clay support could not be completely sub-
tracted in these studies because of the strong Bragg re-
flections from the support itself. Bragg reflections are
known to contribute to the dark field signal in crystalline
materials and to decrease the effectiveness of the Z-
contrast technique (168).

A prereduced Ru/APM catalyst sample was examined at
Arizona State University on a VG HBS STEM to further
explore Ru microcrystallite location and nature. Figure 40
exhibits a high resolution micrograph along with nano-
diffraction patterns from the Ru crystallites indicated.
The dark circle present in the diffraction patterns and in
the bright field image results from a small mirror in the
optical system which is capable of obtaining diffraction
patterns of nanometer-sized crystals (170). The bright
rectangle (BF image) is caused by the beam locater. The
patterns obtained reveal that the microcrystallites were
indeed crystalline, in agreement with previous work for
Ru/SiOz catalysts (171). Figures 40b and 40c show pseudo-
hexagonal diffraction patterns, similar to those obtained
previously for Ru metal <011> and <121> zones, respectively
(171). Although the diffuseness of the Ru aggregates in
the image suggests that they may be located between clay

137

Figure 40. (a) STEM bright field image for prereduced APM
following FT catalysis. Nanodiffraction patterns (b) and
(c) are from the microcrystallites indicated. (Provided
courtesy of J. M. Cowley, Arizona State University.

 

 

138

40

 

139

layers, the quality of small particle imaging is dependent
on many factors, including defocus. The repeat distance of
the fringes in the lower right in Figure 40a is approx-
imately'l.3 nm. These fringes could be assigned as Moire
fringes, yet their general agreement with.the 1.8 nm XRD
spacing of bulk APM samples prior to catalysis suggests
that they are pillared clay-lattice fringes instead. The
magnification of Figure 40a was calibrated using experimen-
tal observation of carbon fringes. The fact that diffrac-
tion patterns corresponded to crystalline Ru metal is
interesting as apparently a pure metal phase is formed by

migration not an oxide or an aluminate.

E. Atypical Features of Ru/APM Fischer-Tropsch Catalysts

The high proportion of branched hydrocarbons and
internal olefins produced by prereduced Ru/APM are two
aspects of these catalysts that are uncommon in Fischer-
Tropsch chemistry.

Both of these anomalies are readily evident in the gas
chromatograms (GC) obtained from catalytic runs. Represen-
tative GC traces are shown in Figure 41 for the gaseous and
trapped liquid fractions. The chromatogram from the
gaseous sample illustrates that branched hydrocarbons domi-
nate the C5+ hydrocarbons, although branching is also
present in C4 products. Production of internal olefins is
greatest for C4 hydrocarbons. The chromatogram of the

trapped liquid demonstrates internal olefin production for

 

140

l I "‘ II "I HI I "III ‘llllllll llllllll III | I

c c
C" 0" ‘7 C. c c10 c11 12 13 c14

‘ II |||||||||||l||||| lllllll lllll ||||

Figure 41. Representative gas chromatograms for prere-
duced Ru/APM catalysts showing production of branched
(Long bars) and internal olefin (short bars) products.
Chromatogram (a) is for the gaseous fraction, and (b)
illustrates the components in the condensed liquid.
Arrows indicate attenuation changes. (250°C, 190 psi,
2930 hr'l, 200 min).

141

up to C10 hydrocarbons. The relative concentration of
internal olefins is inversely related to branched hydrocar-
bon production as they decrease with.higher hydrocarbon
molecular weight.

The high production of methane and other light
hydrocarbons relative to other components is evident by
taking into account the attenuation changes used to keep
the scale expanded. Additionally, hydrocarbon production
to C13 should be pointed out in these chromatograms along
with the relatively large MeOH production detected from the
trapped liquid.

Figure 42 shows the variation of branched/straight
chain (BCn/Scn) ratios on Ru/APM catalysts as a function of
temperature. The production of branched hydrocarbons is
clearly favored with higher reaction temperatures for
Ru/APM although the relationship is not linear. The non-
linearity suggests that this increased branched hydrocarbon
production is affected by more than just the reaction
temperature, as a direct relationship would show a simple
activation energy dependence.

Branched hydrocarbon production in FT typically is
minor compared to production of straight chain hydro-
carbons. However, high pressure and low temperature favor
branched hydrocarbon production on Co, Fe and Ni catalysts.
In Schulzfls (172) recent review on FT, he illustrated that
branching selectivities were greatest for C5 hydrocarbons

on these metals and the product distribution for Ni showed

142

T(°C)
275 250 225 200

 

 

B Cn/S Cn
E3
} ' l
j/
01>

 

 

 

1.01
‘o
1 O A
G
1.8 1.9 2.0 2.1

1/T(K) x 103

Figure 42. Variation of the branched/straight chain hydro-
carbon selectivities with temperature for Ru/APM catalysts.
The symbols represent C4 0 , C5 [3 , Ca A and C7 Q
hydrocarbon fractions. The FT conditions were; 120 psi.,
1910 hr’l, 82/00 = 2, 50 min. TOS, prereduced.

143

30 mol.% branched Cnfor Ni, and 12% for Fe. At these
conditions branching can be very selective, occuring only
in the 2 and 3 positions. This is probably due to steric
demands for branching of small species which strongly
favors fixation at terminal hydrocarbon positions.

Reports of hydrocarbon branching for supported
ruthenium catalysts have been rare, and when branched prod-
ucts were observed.(173) they were in much smaller concen-
trations than their straight chain counterparts. Excep-
tions to this have been observed for Ru dispersed on
silica/alumina mixtures (165) and zeolite supports (165,
174-176). The proportion of branched hydrocarbons however,
has not always been observed to increase with temperature
as a variety of zeolite supported Ru catalysts showed
isobutane selectivity to decrease with temperature (174).

Increases in reaction pressure are commonly associated
with increases in C0 conversion. To separate the effects
that conversion may have on the branched.hydrocarbon yields
BCn/SCn ratios for a pressure study of Ru/APM were norm-
alized with respect to conversion. The changes in this
ratio for the numerous hydrocarbons produced are illus-
trated in Figure 43. The curves obtained demonstrate that
branched hydrocarbons decreased relative to straight chain
Cn with increasing pressure. This decrease was greatest
between 15 psi (1 atm) and 70 psi, after which Ben/SCn were
less sensitive to pressure variation. Maximum proportions

of branched hydrocarbons were obtained in all cases for C7

144

 

E
'3 I3
3 10.0-
>
G I
O
o d
2
_. a D
:3 /
. E] A
3. A/0\A
U
: ‘///// \\\\\f3
1.0«
/o o

A

A

A

L

 

f a

V V V v

3 4 5 6 7 8 9

 

 

q
1

Carbon number (n)

Figure 43. Change in the branched/straight hydrocarbon
ratio ,normalized for conversion variation,at various
pressures. Here thetj represent data at 1 atm.,

A at 70 psi , and Q at 190 psi. Intermediate
pressures follow these trends but are excluded for clarity.

145

hydrocarbons.

As branched hydrocarbon production reported for sup-
ported Ru catalysts has been sparse, the trend toward de-
creased branched hydrocarbon production with increasing
pressure may be characteristic for APM as a Ru support.
Increased pressures would increase the concentration of the
intermediate responsible for branched hydrocarbon produc-
tion, yet it is plausible that these intermediates may be
consumed by a competing reaction favored at elevated pres-
sures.

The propensity for production of C7 branched hydrocar-
bons compared to lighter and heavier fractions is of in-
terest as industrial reforming reactions strive to achieve
high proportions of branched products in the gasoline range
due to increased octane ratings. An examination of longer
run times reveals that this selectivity is maintained
(Figures 44 and 45). Comparison with the first Ben/SCn
versus 1/temperature plot (Figure 42) shows that although
these ratios decrease with carbon number, the greatest
branched/straight chain ratios are obtained for C7 and C6
hydrocarbons. These figures demonstrate that the selecti-
vity of C7 and C6 branched hydrocarbons is stable with
longer run times and represents a significant departure for
Ru/APM catalysts compared to typical supported Ru cata-
lysts. It should also be noted that isobutane fractions
decrease with higher temperatures at longer reaction times

contrary to the results at 50 minutes. This relationship

146

T(°C)

275 250 225 200

 

D\.
./
/

EJOI>

O

 

 

 

V V V

1.8 1.9 2.0 2.1
l/T(K) x 103

Figure 44. Production of Branched/ straight hydrocarbons
at longer run times (100 min.) for 04 C) , C5 E]

Co A , and C7 C) hydrocarbons. Pressure 1 atm.

Run conditions were the same as in Figure 42.

B Cn/S Cu

147

 

 

 

 

 

T(°C)
275 125() 2255 2(10
J
10.0 - 0 Q
A
4 Q
1 \A
E]
1.0‘
T O/M
1.8 1:9 2:0 2:1

1/T(K) x 103

Figure 45. Steady state (200 min.) selectivity for
branched hydrocarbons for C4 0 , Cs B , Co A , and 070
products. Run conditions were the same as in Figure 42.

148

is consistent with the earlier mentioned trend for
Ru/zeolites (174) that originally appeared contradictory to
Ru/APM results.

The fact that branched hydrocarbons are not normally
produced in FT has led several authors (165, 174-176) to
propose that the acidic nature of supports such as
silica alumina mixtures or zeolites are responsible for
hydrocarbon branching. King (165) found that a mechanical
mixture of Ru/A1203 with an ultrastable zeolite gave prod-
ucts rich in isobutane and isopentane, whereas the Ru/A1203
catalyst alone gave only minor branched products. This
strongly suggests that isomerization of primary FT hydro-
carbons occurs subsequent to initial FT production and that
this could take place downstream of the original active
site. If this is the case the production of secondary,
branched products should be affected by the feed gas
velocity.

Figure 46 shows the result of a study on Ben/sen
ratios for Ru/APM catalysts run at different flow rates.
The Ru/APM catalyst was run at the gas hourly space veloc-
ity (GHSV, V/V/h) of 2700 hr'1 for 3 hrs at 275°C. This
allowed steady-state hydrocarbon production to be achieved
before the flow was increased and equilibrated at each flow
rate for 45 min. .After the data points were collected for
various flow rates the flow was returned to the original
values and equilibrated for 45 min so that an estimate of

the effect of cycling could be obtained. The decreasing

149

 

10.0 ¢\

B Cn/S Cn
E:

1.0 1

 

 

 

2700 5400 8100 10,800

Space Velocity (hr‘1)

Figure 46. Effect of space velocity on the branched/
straight ratio. The Ca hydrocarbons are represented by(3 ,
C5 by [3 , Cs by [5 , and C? are shown by C) . The

slashed symbols show the ratio when the flow was returned to
. the original value. (275°C, 110 psi, 82/00 = 2, 45 min.
TOS, prereduced, contact time from 1.3-0.33 seconds).

150

production of branched relative to straight chain products
with increasing flow rate is clearly evident. The magni-
tude of this effect was greatest for the heavier hydroc-
arbons (C6, C7) and decreased for lighter branched pro-
ducts.

A number of factors may influence the Ben/SCn ratios
as the flow rate is increased including: conversion,
residence time and diffusivity of individual hydrocarbon
components in the microporous APM, flux of product mole-
cules at the catalyst surface and location of sites respon-
sible for isomerization. We would be remiss at this point
if we did not point out the complexity associated with
these factors, which are coupled with the complicated
kinetics, and mechanism of FT. .A simple, yet plausible,
explanation would be that the decreasing diffusivities of
higher molecular weight hydrocarbons would inhibit diffu-
sion of molecules such as 1-heptene back into micropores
once it has left the APM interior, relative to say 1-
butene. Increasing flow rates would magnify such differ-
ences as the alkene intermediates are pushed through the
catalyst bed with a higher velocity and reaction of mole-
cules with lower diffusivities would.become diffusion
limited.

The relationship between the ratio of internal olefins
to terminal olefins with reaction temperature is shown in
Figure 47. This ratio decreases with increasing tempera-

ture as does the ratio of internal olefins/straight chain

151

 

T(°C)
275 250 225 290
10.0 L ‘ ‘
d
I /@
.1 [b/

l Cn=/ Cn-
0\

1.0

 

O
O

i Cn=/T Cu:

 

 

 

1.8 1.9 2.0 2.1
l/T(K) x 103

Figure 47. Production of internal olefins relative to
terminal olefins (i Cn=/T Cn=),and n-alkanes (1 Co‘/ Co‘)
as a function of temperature. Here the symbols are defined
as;iC¢=/'I'Cq= is Q , iC4=/Ca' O , iCs‘l‘l‘Cs’ A , and
iCs‘le‘ is [j . Conditions were 120 psi. , 1910 hr’l,
Hz/CO=2, 50 min. TOS, prereduced.

152

hydrocarbons. The proportionally smaller production of 05
internal olefins is also evident from this plot. Kellner
and Bell (173) found that olefins exhibited a higher acti-
vation energy for formation than paraffins (about 6 kcal/-
mole), indicating greater olefin production should occur
with elevated temperatures. Their experiments revealed
however, that above 215°C the olefin to paraffin ratio
decreased, and they attributed this to increased
hydrogenation. This would also explain the observed
decrease with Ru/APM catalysts.

The internal olefin production versus terminal olefin
hydrocarbon ratios were essentially invariant with respect
to pressure. When the sum of all alkenes (including
branched olefins) was compared with the total alkane prod-
uction and normalized for the different conversions
obtained with the pressure study, the plot of Figure 48 was
obtained. The relative position of these curves suggests
that the total olefin yield relative to paraffins generally
increased with pressure. .Although this might be due to the
increased conversion obtained with greater pressure these
plots were normalized with respect to conversion and sepa-
rate plots of alkene/alkanes versus conversion did not show
a significant correlation between these two parameters.
The dip in Cng/Cn' for C4'C6 hydrocarbons may be due to
enhanced olefin hydrogenation with the longer residence
times expected for higher molecular weight hydrocarbons.

The increase beyond C6 hydrocarbons however is not

153

 

1.0

\<
\

 

 

 

x ‘ C] cy———-C>
A
7‘ ‘ \A
6'
'\ 4
3‘: O
.01.
1 G
1
‘4 5 6 ‘7 8

Carbon number (n)

Figure 48. Comparison of total alkene/alkene , normalized
with respect to conversion, for 1 atm. O , 70 psi A , and
190 psi [3 . FT conditions were; 250°C, 1910 hr'l, Hz/CO=2,
50 min. TOS, prereduced.

154

explainable by this reasoning.

The ratio of alkenes to alkanes determined from the
flow rate study, previously described for branching, is
given in Figure 49. The trends for C4,C5and C6 hydro-
carbons all show increased olefin production with increased
space velocity. Yet, the trends are not smooth and a few
deviations from these trends are evident. A return to the
original flow rate followed by 45 min of equilibration
showed only slightly higher Cng/Cn' ratios than obtained
originally. A breakdown of these olefins into icn'/Tcn=
and icn‘/ Cn- ratios is given in Figure 50. The icng/Tcn=
ratios do not correlate well with the increased flow rates
but the icn=/Cn' ratios for C5 and C4 do exhibit a regular
dependence on space velocity.

The dependence of Cn'/Cn' on flow rate is closely re
lated however to conversion, as conversion also decreased
steadily with increased flow rates for Ru/APM catalysts.
This dependence is supported by previous studies on
supported Ru catalysts (165,174,177) which all showed the
Cn'/Cn' ratio to decrease with increased conversion. ‘With
increased conversion greater incorporation of light olefins

into growing hydrocarbon chains is expected.

F. Isomerization Mechanisms

The acidity of APM clays is well documented, and the
involvement of this acidity in isomerization of FT products
to branched paraffins and internal olefins is supported by

studies relating acidity to these unusual features. Olefin

155

 

10.0‘
C)
. G)/ \
1.0« Q// 51/8

Cn‘/ Cn'
R
//>
D’

051.

 

 

 

2700 5400 8100 10:800

Space Velocity (hr‘l)

Figure 49. Ratio of alkenes to alkanes for Ru/APH catalysts
as a function of flow rate. The symbols show 04 (D ,

Cs E and Cs (3 hydrocarbons. The slashed symbols show the
recycle values. Conditions were 275°C, 110 psi., Hz/CO=2,
45 nin. TOS, contact time from l.3-0.33 seconds.

156

 

10.01 A
‘ / \A
\A
- <3)

I: l Cn=/ Cu"
[30

G E]
.. 1.0-
5 .
5" 1
\
:3 .
-H

 

 

 

27'00 54100 81'00 10,800

Space Velocity (hr‘l)

Figure 50. Yields of internal olefins relative to

l-alkenes and n-alkanes as a function of flow rate. Symbols
are; iCq=/1-C4= Q , iC4=/Cn' Q) , i C.r.=/l-Cs= A

and iCs=/Cn’ E] (275°C, 110 psi., Hz/CO=2,

45 nin. TOS, l.3-0.33 sec. contact time).

157

isomerization reactions occur readily on acid catalysts.
The relative strength of acid sites required for these
carbonium ion reactions increases in this order: cis-trans
isomerization < branching < alkylation (12). Skeletal
isomerization of alkanes is more difficult than that of
corresponding alkenes and high temperatures or strongly
acidic sites are needed, but the presence of traces of
olefins markedly accelerates this reaction as they are
readily protonated to form carbonium ions (13).

The relevance of support acidity was indicated by
King's study (165) where it was determined that the Si/Al
ratio of silica-alumina mixtures and zeolite supports ap-
peared to be important in affecting enhanced yields of
branched products. Results from Chen gt a}; (175) added
additional support for the influence of Si/Al ratios on
branching selectivities when zeolites were exchanged with
Ru(NH3)63+. Yet when these same zeolites were prepared
using vapor impregnation of ruthenium carbonyls at the
identical dispersion and Si/Al. No isobutane was obtained.
Thus the Si/Al ratio does not directly determine the iso-
butane yield. They suggested that greater Si/Al ratios
increased the acid strength of the OH groups formed within
the zeolite support when the cluster decomposed during cal-
cination (ie. the formation of 3H+/Ru3+). Thus the pres-
ence of these Bronsted acid sites may initiate branching of

primary FT products, creating bifunctional catalyst proper-
ties.

158

It is also plausible that the changing strength of the
Si/Al ratio may induce hydrogen suppression of neighboring
Ru crystallites (178). Nearly a monolayer of hydrogen is
adsorbed on Ru surfaces during catalysis (179) and altera-
tion of this hydrogen concentration could effect chain
growth, hydrogenation and hydrocarbon isomerization.

A number of mechanisms have been proposed to account
for hydrocarbon branching in FT, the most illustrative ones
are depicted in Figure 51. The incorporation of propene in
growing alkyl chains was postulated based on 14C experi-
ments for the normal pressure cobalt synthesis, yet this
has not been supported by additional experiments (71). The
addition of a methyl species into chemisorbed growing
chains, and isomerization of fixed chains has also been
cited as possible mechanisms (172). The formation and
rearrangement of a alkylidiene species has also been sug-
gested as a intermediate responsible for branching (127).
Decreases in hydrogen concentration would enhance chain
branching according to this mechanism. This mechanism was
supported by the finding that K promoted Fe203 catalysts
produced branching whereas unpromoted catalysts gave no
branched hydrocarbons. They also noted that K promotion
increases CO adsorption and decreases H2 adsorption. A
mechanism involving olefin protonation and.rearrangement
through a carbonium intermediate would also account for the
observed results. Although there would be a significant

difference between branching and nonbranching

(s)

(b)

(C)

(d)

(c)

Figure 51.

159

 

 

 

 

 

CE:
I
RC3: 1* 083082132 b RC3: -CH-Cflz
CH:
808 CH: R‘tfi
tun-”war + ”A," ”may”
3 CH:
n\c£caa ‘cf
RCHzCHz C0 2 H: CE:
W —’ ”Ly-y ' 310
R 8
CH: CH:
CH: CH:
CH: -B. C! CH: BC!2¢I:CHCB¢
—_\
.—_ 44)... 4.3.
F
Ila
B'-CH—CB: I
CH: CH: 30308::
t

I
11
(Ln-3
C /.‘Q-°.
_e_c_c_a' l——. -C-C-h- R’

Plausible mechanisms to account for hydrocarbon

branching showing: (a) insertion of propene, (b)addition of
chenisorbed nethyl species, (c) isomerization associated
with hydride shift, (d) formation and rearrangement of an

alkylidiene,

(e) carbocation rearrangements .

160

rearrangements in such a mechanism the high acid strength
in clay interlayers (chapter I) and the high concentration
of acid sites within micropores could account for the
branching propensity of APM. The isomerization of terminal
olefins could take place on these acid sites also as illus-

trated below:

CH8 /R .
‘ + H . Q I C=C ois
;c:c--c—R ‘__—"" -C—C-C—-R /' / \
I ._ H I 1 1 \ /R
)=C trans
CH3 \

Silica is known to be a good isomerization catalyst and
experiments with precipitated iron on silica catalysts
showed that as the silica content in a catalyst series in-
creased so did the production of internal olefins relative
to l-olefins (180). This was interpreted as evidence that
most of the internal olefins were produced by isomerization
of l-olefins.

Based on the relevance of these mechanistic arguments
to the selectivity of branched and isomerized products on
Ru/APM, additional experiments were conducted with the
addition of alkene probe molecules, and variation of Hz/CO

ratio.

G. Alkene Probe Addition, and Hz/CO /Variation.

In the first experiment a known concentration of CZ'CG
olefins (in He) was added to a Ru/APM catalyst that had
been run in FT synthesis for 3 hrs at 120 psi and 275 °C to
achieve steady state conditions. The relationship between

the masses of the alkene standard added was 02‘ 4.2, C3=

161

6.5, C4= 9.3, 05= .9, 05= 1.0. The concentration of
alkenes added in the first step of this experiment was
roughly similar to that of total hydrocarbons produced for
each Cn and the second concentration corresponded to twice
this concentration. To achieve this higher concentration
the flow of alkenes relative to C0 feed was altered (total
flow kept constant.

Figure 52 shows the distribution of hydrocarbon prod-
ucts detected relative to methane for the FT production
before alkene additons, along with those after 40 min of
eqilibration with each alkene concentration. The distribu-
tions obtained form the alkene additions parallel each
other closely, but some differences are evident relative to
the initial FT distribution. Even though the concentration
of alkenes added increased as C4=> C3: > C23, the concen-
tration of C4: >> (05', 06‘) should give greatly decreased
hydrocarbon production for > C4 hydrocarbons. Yet, 05 and
C6 hydrocarbons continued to increase relative to the
steady state FT distribution shows that production of C7,
C8 and C9 hydrocarbons was also increased by these alkene
additions.

Figure 53 indicates the mass of olefins incorporated
for branched and straight chain hydrocarbons relative again
to the steady state SF production. From this plot it is
clear that straight chain hydrocarbons showed increasing
production in the order C6 > C5 > c4 , but production of

hydrocarbons > C6 actually decreased slightly for both

162

 

‘ ..- . A!

100.0. \

./
/
f

/
4.
/

W Cn/ H CH4

 

 

 

2 31 4 5 6 7' 8 9

Carbon number (n)

Figure 52. Increase in hydrocarbon production resulting
from incorporation of l-alkene probe molecules. The symbols
signify; steady state FT (100 min.) C), hydrocarbon dis-
tribution during the first alkene addition A , and after
the second additon [j . The [alkene] in the second add-
ition was twice the first. (275°C,llOpsi.,4336 hr’l,

Hz/CO=2, prereduced).

163

 

A-ALA

10000:

‘R ‘943

(nanograms/ml of gas)
/0 s

100.01 ‘ ‘

I: 8 Cu
if

A

30::

 

 

J

t v j

4 5 6 7 8

Carbon number (n)

‘01
.5
0

Figure 53. Production of branched and straight hydrocarbons
before and after l-alkene additions. The 43 show steady
state FT (180 min.), and G and B production after the
first and second alkene additions, respectively. The slashed
points are for straight isomers and the regular symbols are
for branched products. (275°C, 110 psi., 4336 hr",

Hz/C0=2, prereduced).

164

alkene concentrations. Enhanced production of 06 and C5
branched products were similar and were greater than that
for C4 branched products. Production of branched products
dropped sharply for Cn > C6 but remained somewhat higher
than initial FT production for C7 -C11 products. This
indicates enhanced production of hydrocarbons greater than
the 02-06 olefins added occurred, and that olefins were
incorporated into branched hydrocarbon products. This is
consistent with the incorporation of olefins into growing
chains and the proposal that Ru/APM catalysts are selective
for branched products due to their intracrystalline acidity.

Figure 54 shows a breakdown of olefin incorporation
for the internal alkenes produced in FT with these addit-
ions compared to the paraffin incorporation. The incorpo-
ration of olefins added also is evident in the production
of internal alkenes. Based on the relationship obtained,
which is nearly linear, it is also clear that cis-trans
isomers are also increased for C7 hydrocarbons. This also
lends support to the ability of Ru/APM to isomerize l-
olefins to cis-trans isomers. The decreasing production of
internal olefins with increasing Cn for Ru/APM is however
contrary to the opposite trend observed by Dictor and Bell
(127), who used iron based catalysts. They attributed
their increased yields of internal olefins to longer resi-
dence times for higher Cn' The strong hydrogenation activ-
ity of Ru/APM catalysts however could readily transform

internal olefins to paraffins which would explain this

6: Cu-

. Cu:

165

 

(nanograms/ml of gas)

 

 

 

10000 +7,
1 [A
I a
. fif/A\a
s a“: \A
A
d 0/\““G
.ij (3
[063‘ o\g
1000- ‘ ‘\
'1 ‘ \
. ~. FE
. ‘/R.
I \ \ \
\ \ \
4 . \
1 g\ \F
\ ‘\#
u 9’ L
1
v If r —r v v
3 4 5 6 7 8

Carbon number (n)

Figure 54. Breakdown of the l-olefin probe molecules
added during FT into internal olefins (iCn=), and
total paraffins (Cn‘). The regular symbols for FT
steady state (180 min.) G , first addition A ,and
second 8 refer to paraffins and the slashed symbols
to internal olefins. FT conditions as in Figure 53.

166

discrepancy. The olefin/paraffin ratios obtained for
Ru/APM are not as high as other supported Ru systems which
could be due to this hydrogenation propensity.

When the CZ-Cs olefin mixture was passed over the
Ru/APM at 275°C, 120psi, in the absence of C0 (prior to FT)
essentially 100% of the C2 , C3 olefins were converted to
paraffins, demonstrating this hydrogenation activity. Re-
sidual alkenes for C4, C5 and C6 could just barely be
detected. Chuang gt g}; (81) showed that 97% of C2: added
to Ru/Sioz catalysts under synthesis conditions (C0
present) was hydrogenated to ethane. They detected 2%
incorporation of ethylene into growing chains. The alkene
addition to Ru/APM in the abscence of CO also yielded large
quantities of methane which is due to hydrogenolysis.
Chuang gt gt; (81) did not observe hydrogenolysis of the
added ethylene, but cited references supporting the fact
that this was due to the presence of CO, and competition of
reactions such as chain growth. Hydrogenolysis on Ru has
been studied extensively and the reader is referred
elsewhere (182) for additional information.

To diminish the potential masking effects of hydrogen-
ation in the production of internal olefins and branched
hydrocarbons on Ru/APM catalysts the mass of alkene/mass of
Hz was varied in the absence of C0. H2 /CO ratios were
also varied during FT to examine these features.

For the first experiment the mass of alkene added to

the Ru/APM catalyst at 275°C and 1 atm, was changed with

167

respect to the flow of H2. To obtain higher dilutions of
the H2 in the feed gas He flows were also used. Flow rates
were maintained at constant values throughout these experi-
ments. These alkene additions were separated by a 30 min
period of pure H2 flow.

The relationships shown in Figure 55 for this study
illustrate that the mass of en. produced relative to
methane decreased as the relative concentration of Hz was
decreased. ‘The lines drawn through the points show good
correlations for all Cn- except for C3. These trends
illustrate suppression of the hydrogen concentration at the
metal surface because of the decreased conversion of added
olefins to paraffins. The increase in olefin concentra-
tions with decreased H2 concentration, shown in Figure 56,
is in agreement with the paraffin plot (Figure 55). Of
particular interest is the dramatically increased yields of
both cis and trans isomers with decreased H2 concentra-
tions. The ratio of cis to trans isomers would be expected
to reach equilibrium values when skeletal rearrangement is
occurring (13), whereas if they were produced as primary FT
products their concentrations may be equal. It is also of
great significance that at the lowest concentration of H2
branched products were observed clearly in the GC trace
(Figure 57). Although their concentration is low relative
to that of the neighboring paraffins produced by hydrogena-
tion, they are similar in concentration to the peaks that

represent the residual alkenes. This clearly demonstrates

168

 

g

~ 1

x 4 A A
5

° ‘101J1. £5 £5

3 I

8 q o

3 /\

“~\\\\\\\‘Ej

. . \ O
1 \ \ A
0

V r

0 1 2 3
W Alkene/ W Hz

\0\ a
‘ \,-.\ \m
o

 

 

 

4

Figure 55. Transformation of l-alkene probe molecules
to alkanes in the absence of CO (no FT) as a function
of the alkene to hydrogen ratio. The symbols represent;
02 0,0: A ,c. /‘\ .05 C) ,and c. [:1 . (275°C, 1 atm.,
3230 hr‘l)

169

 

1013

A1111

\1
\\

1.0 / fl/fi A

102

J>\
O I
\
- O

051

W (1 Cn=/ CH4)
D
0\

LIAAJA

 

 

 

W (l-Alkene/ H2)

Figure 56. Transformation of the added l-alkene probes
to internal olefins as a function of the alkene to ,
hydrogen ratio. The symbols are; cisC4 O , trans C4 0 ,
cis Cs A , trans Cs A , l-C4= O ,l-Cs= /°\ /
(275°C, 1 atm., 3230 hr'l, prereduced.

170

 

 

 

'6‘”? ID Hm
u L! U U U
# 4 J- # +
a
... H
u
1
In: I '(D
b \J U” 1am U
V
V V
H [II II
c ' v
' v
'11 III II
Figure 57. Gas chromatograms of l—alkene probe molecules

after passing through the Ru/APM catalyst bed. (a)25°C
before any isomerizations, (b) reactor effluent when alkenes
are added to the catalyst at 275°C , 1 atm., 3230 hr‘l,

50 min. TOS,(1-Cn=/H2=l.25) ,(c) l-Cn=/Hz=2.8. Long bars
indicate branched products and short bars internal olefins.

171

that l-olefins can diffuse through the micropores of APM
and be isomerized to branched products as well as internal
olefins. Further, desorption of branched products through
the micropores is possible for C5 and C6 hydrocarbons. The
propensity for methane formation through hydrogenolysis (in
the absence of C0) is also evident from the GC trace. The
low H2 concentration required for this branching to be
detected suggests that H2 concentration is very important
to branching.

To examine the effects of decreased hydrogenation on
branched hydrocarbons and internal olefin yields the 82/00
ratio was varied during FT synthesis conditions. This was
done following the alkene additions above by halting the
alkene addition and increasing the H2 flow to 35 ml/min for
30 min to flush residual alkenes. The pressure was then
increased from 15 to 120 psi while the H2 flow was kept
constant. After 45 min at this pressure and 275°C the CO
was added, and the H2 decreased so that a total flow rate
of 35 ml/min was obtained. Fischer-Tropsch hydrocarbon
production was equilibrated for 55 min before the first
data point was taken. IEach subsequent Hz/Co data point was
taken after 50 min of equilibration.

Figure 58 relates the data obtained for the change in
BCn/SCn with Hz/CO ratio. The relationships obtained show
that branched hydrocarbon production increased with de-
creased 112/CO. This was consistent with the data obtained

for the alkene additions. Figure 59 further demonstrates

172

 

on (0
0\
/o
0

N
\
/

/7

B Cn/ n Cn‘
(a) A
b E]

0
//;7//
O

 

 

 

Hz / CO

Figure 58. Change in the selectivity for branched hydro-
carbons during FT with Hz/CO. The symbols are; C4 C) ,
Cs A .Cs E] ,C? Q .08 /'\. (2750c, 120 psi.,

3230 hr‘l, prereducedL

173

 

 

 

 

 

 

s
43

0

s
'0

o

a

an 9 ‘

I:

o

s 3 A:

a 1 “\A
II

B

O 7 4
5

E]

3

w 6‘

0H

h

o

I

o

.2 5

s

o

z
2 4

2:

0" B
a 3 O

s:
c, 2.
'8

w C)
'H

23 1 1

I

o

s

H i

(15 1 2

82/00

Figure 59. Isomerized hydrocarbon production 1;; normal

FT production (n-alkanes and l-alkenes) as a function of the
82/00 ratio. Here C4 hydrocarbons are shown as C) ,

Cs as E] , and Cs as A . (275°C, 120 psi., 3230hr‘1)

174

that decreased H2 resulted in increased mass of isomerized
products (branched hydrocarbons and internal olefins) rela-
tive to initial FT products (terminal olefins, and straight
Cn)°

These data clearly support the argument that decreased
hydrogen concentration will result in increased yields of
isomerized products. The correlation of enhanced branched
hydrocarbon production with decreasing [H2] also supports
the argument that 1-olefins are isomerized to branched hy-
drocarbons as their concentration would increase with lower
Hz/Co. The increased olefin concentrations at lower Hz/CO
would allow a higher proportion of molecules to reach the
Bronsted acid sites of Ru/APM and be isomerized into
branched products.

The isomerization selectivity for Ru/APM catalysts,
due to the interaction of primary FT terminal olefin prod-
ucts with APM intracrystalline acid sites, is enhanced for
prereduced catalysts as the Ru is retained within APM
‘micro- and mesopores. Terminal olefins that originate on
these Ru microcrystallites must diffuse past numerous
intracrystalline acid sites on their way to the APM
exterior. These interactions are strongly enhanced by the
retention of Ru within the pore structure, which results in
prereduced catalysts having a much higher selectivity for
isomerized products than their nonprereduced counterparts
(external dispersion of Ru). The chemical nature of APM

then influences the FT production of hydrocarbons on metals

175

dispersed in pillared clays, which results in this unique

isomerization selectivity for Ru.

H. Conclusions

The method of highly dispersing ruthenium in the
micropores of alumina pillared montmorillonite (APM) by the
intracrystalline protonation of Ru3(C0)12 presents poten-
tial advantages for heterogeneous catalysis. The ruthenium
is dispersed within the micropores in a low oxidation state
for example, and a prereduction in Hz was not required for
activity in the hydrogenation of carbon monoxide (FT).
Catalytic characterization of Ru/APH showed that deviations
from SF were produced at short reaction times such as the
apparent existence of more than one chain growth probabili-
ty. Such deviations were short lived however, and after
one hour hydrocarbon distributions closely followed SF.
Catalysts that were prereduced in hydrogen conversely ex-
hibited no initial SF deviations, but did give higher
conversions and favored the production of higher molecular
weight hydrocarbons.

High resolution electron microscopy studies proved
vital in investigation of differences between nonprereduced
and prereduced catalysts. These investigations revealed
that in nonprereduced catalysts considerable Ru sintering
occurred. The Ru migrated from the micropores of these
materials and formed large crystallites on the.APH support

during catalysis. Nonprereduced catalysts run 30 hrs in FT

176

had a broad distribution of external metal crystallite
sizes with an surface average mean particle size of 56 nm.
Prereduced Ru/APM catalysts were significantly different as
the majority of Ru atoms were retained within the.APM
during catalysis. The distribution of metal microcrystal-
lites was much narrower for these catalysts and crystal-
lites greater than 10 nm were not detected in the numerous
samples studied by EM. Catalysts run for 30 hrs. gave a
mean particle size of 6 nm, nearly an order of magnitude
lower than their nonprereduced counterparts. Investigation
of Ru/APM catalysts after reaction showed thatmicrocrys-
tallites on the order of 5 nm gave nanodiffraction patterns
for Ru metal not ruthenium aluminates or oxides.

A model was proposed for the nature of the clay sup-
port to account for the presence of Ru microcrystallites on
the order of 5 nm as the gallery height of 1 nm for APM
should not accomodate such large crystallites. This model
shows that such crystallites can be held in layer packing
disorders created when clay layers stack atop one another
in the air drying process. Physical characterization of
APM lent support for this model, and the application of z-
contrast imaging showed that indeed.Ru microcrystallites
were present in thicker regions where such void spaces
would be numerous. These void spaces may trap Ru micro-
crystallites formed during the prereduction step and hinder
their migration to form larger crystallites. INonprereduced

catalysts conversely may interact with C0 present in the

177

feed gas when FT begins, forming molecular clusters which
readily desorb from the micropore regions and sinter on the
APM surface.

Further catalytic examination of prereduced catalysts
revealed that the carbon number distributions produced in
FT followed SF regardless of run conditions and the con-
tainment of Ru within a micropore cavity did not give
anomalous distributions of Cn.

The two features that are surprising for Ru/APM cata-
lysts in FT though are the high selectivities for branched
hydrocarbons and internal olefins. Branching was found to
be dependent on temperature, flow rate and to a lesser
extent pressure. Internal olefin production was also
dependent on these parameters, and these atypical products
are changed somewhat with alteration of the C0 conversion.

As production of these fractions is quite unusual for
supported Ru catalysts potential mechanisms were discussed.
These mechanisms pointed to the intracrystalline acidity of
APM playing a significant role in production of these
uncommon isomers. .Addition of l-olefin.probe molecules
showed that olefins were incorporated into higher molecular
weight olefins and paraffins in agreement with previous
literature results. ‘Yet, these experiments also showed
that these olefin probes were incorporated into branched
hydrocarbons and internal olefins. Variation of the alkene
/H2 ratio showed that decreased hydrogen concentrations

increased the yield of internal olefins and provided

178

evidence that l-olefins could diffuse to the acid sites and
be isomerized to branched products as well as internal
olefins. A study of production of these unusual products
as a function of Hz/Co ratio during FT catalysis proved
that production of branched hydrocarbons and internal
olefins increased with decreasing H2 concentration. These
experiments with alkene additions and H2/00 variation added
considerable support to the proposal that these products
have enhanced selectivities for Ru/APM catalysts due to the

intracrystalline acidity of alumina pillared clay.

APPENDICES

Appendix A: Minimization of Microscope Artifacts

Ru/APM prior to catalysis contained essentially tria-
tomic ruthenium centers weakly bound in the silicate inter-
layer, as shown by the physical characterization methods
mentioned in Chapter Iva However, initial STEM studies of
the Ru/APM prior to use as a FT catalyst showed no discer-
nible Ru clusters when the sample was mounted on an ambient
temperature stage. These triatomic clusters are apparently
too small to scatter sufficient electrons, relative to the
clay support, to be detectable in bright or dark field
images. After extended periods (hrs.) of exposure to the
electron beam however, the sample exhibited Ru aggregates
on clay particles microns from the areas initially studied
(159). Large aggregates (> 50 nm) exhibited symmetrical
shapes, such as the pseudo-hexagonal and rectangular aggre-
gates illustrated in Figure 60. These particular aggre-
gates gave well ordered microdiffraction patterns consis-
tent with hexagonal and monoclinic lattices as shown in
Figures 60b and 6°C, respectively. The presence of large
Ru aggregates was inconsistent with the initial STEM
studies and physical data mentioned earlier, so several
regions of the sample were examined on numerous squares of

the Cu grid. Ru aggregation appeared to diminish on

179

180

Figure 60. STEM dark field image of Ru crystallites
formed by beam induced agglomeration in the micro-
scope. Microdiffraction patterns are from the
pseudo-hexagonal (a), and rectangular (b) crystal-
lites. Sample stage at room temperature.

 

182

surfaces far from the initial study areas, but in some
cases Ru aggregation was significant. The Ru appeared to
cluster on external clay surfaces near edges or other sharp
features that may have acted as nucleation centers.

Atom migration due to localized specimen heating has
been reported previously (171,183-85). The selective loss
of low mass elements or weakly-bound species due to lo-
calized beam heating also has been observed (183,184).

When a liquid nitrogen stage was used in the micro-
scope, transported Ru aggregates were not found and Ru
migration was minimized effectively'in.the numerous samples
examined. .Although the mechanism of Ru migration warrants
further study, the important point is that with the use of
a cold stage beam-induced Ru agglomeration was not evident.
Thus, the Ru crystallite sizes observed for the spent cata-
lyst should be truly reflective of the size distributions
resulting from catalysis. It is noteworthy, however, that
when the Ru/APM catalyst precursor was investigated by EDS
microanalysis at 143 K with a tilt stage, characteristic Ru
x-rays were not observed. It may be that the Ru carbonyl
species on the surface were boiled-off by the beam or that
the low Ru loadings of these catalysts (0.2-0.4 wt%) were

near the EDS detection limit.

Appendix B: Feed Gas Impurities

The results from Chapter IV were obtained on catalyst
samples exposed to purified reagent gases. In control runs
for iron pillared clays catalytic characterization of the
activity of the reactor system (mainly CH4 production) was
negligible relative to the hydrocarbon production from the
Fe/PILC. With the low loadings of the Ru/APM system, and
low conversion for nonprereduced catalysts, this background
activity was not negligible.

Analytical STEM studies revealed the dramatic conse-
quences of the use of technical grade gases. Figure 61
shows nonprereduced Ru/APM catalyst samples after 30 hrs of
catalytic reaction in unpurified H2 and C0. EDS analyses
of these crystallites, shown in Figure 62, proved that not
only Ru but also Ru/Fe and Ru/Ni alloy aggregates were
present on the clay. The large concentrations of these
impurities relative to Ru explains the high apparent metal
concentration.

Numerous experiments proved that the contaminants were
originating from metal carbonyls present in the feed gas,
which was purchased in a steel tank; By changing the C0
tank from steel to aluminum, use of ultra high purity
gases, and the through purification procedures outlined in

183

184

Figure 61. Hu/APH after 30 hrs. of FT reaction with
feed gas containing carbonyl impurities at low (a),
and high magnification (b). Microscope sample stage
at room temperature.

185

 

COUITS

 

186

£1

1 74 nm Ru/Fe Aggregate

 

 

COIITS

 

 

30 nm Fe/Ni Aggregate

 

v v v v V v v a V f v v v v v f

600.18

 

IIIIOV (he!)

6 nm Ru/Fe Aggregate

s a
I a
A A

v v v v v

—
fl 1 fi 1 V V fl
0 5.. I... I... :0)

IIIIGV (ie.)

Figure 62. EDS analyses of the aggregates indicated in
Figure 61, which show the presence of Fe and Ni impurities.

187

Chapter II this problem was eliminated. Control reactions
of the new reactor system , run > 10 hrs., revealed that no
background activity from the reactor was obtained up to 180
psi. when the reactor was reduced in H2 at 420°C overnight.
The lack of metal carbonyl decomposition products inside
the quartz insert further confirmed the successful removal
of metal carbonyls. Finally, EDS microanalysis of post-run
catalysts in the STEM did not show Fe, or Ni present on the

catalyst surface.

Appendix C: Preparation of Oxygen Scrubber

The synthetic method of Mo Ilwrick and Phillips (105)
was the basis for the preparation of the Mn (II) oxide/
Sioz oxygen scrubber used in the present work. They
reported that Mn (II) oxide/ Celite diminished the oxygen
concentration in N2 streams to the parts per billion range.

The impregnation of Mn (II) on Si02 began by dis-
solving 5009 of high purity Mn(C2H302)2 x 4H20 (Riedel-
DeHaen AG) in enough MeOH to cover one pound of 12-28 mesh
SiOZ (Aldrich) placed in a crystallizing dish. The solu-
tion was dried by gentle warming with a hot plate (oven
drying at 50°C would be preferable) with occasional stir-
ring. After transferring the material to a baking dish
and heating in an oven at 70°C overnight the solid was
placed in calcining tubes. Argon was passed through the
tubes (hood ventalation required) and the temperature was
slowly increased to 400°C , which was maintained for 5 hrs.
A vacuum was then pulled on the material at 400°C to
insure complete removal of acetic acid. After slowly
oxidizing at 25°C, air was passed over the material at
410°C, and cooled before it was resieved.

Nitric acid (3N) was added to the solid in a

188

crystallizing dish to remove any traces of other metal
oxides, and to assure its catalytic inactivity. After 30
min. it was filtered with a Buchner funnel. It was then
washed with copious quantities of deionized water. The
Mn(II) oxide/Si02 was dried at 410° under vacuum before
being added to the reactor system. This catalyst was
activated in H2 (50 ml/min.) by very slowly increasing the
temperature to 420°C and maintaining this temperature over-
night. The catalyst was tested for response to 02 and
reacted instantaneously when exposed to oxygen, turning
from green to black.

This catalyst was also synthesized using mangenous
nitrate (50wt% solution, Fischer) and Sioz beads 2-4 mm in
diameter (Sigma), that had been additionally purified by
acid and washings with deionized H20. The steps for
impregnation were similar, but the nitrous oxides that
evolve during heating required the use of a hood in an
unoccupied room. This material was somewhat more difficult
to reduce and was used only where larger catalyst particles

were required.

189

LIST OF REFERENCES

10.

ll.

12.

13.

14.

REFERENCES

Davies, J.E.D.: Kemula, W.: Powell, H.M.: Smith, N.O.
9:. Incl. Phenom. 1983, _1_, 3.

 

Setton, R.; ed. "Chemical Reactions in Organic and
Inorganic Constrained Systems," D. Reidel, New York,
1986.

Whittingham, M.S., "Intercalation Chemistry,"
Whittingham, M.S.: Jacobson, A.J., eds., Academic
Press, New York, 1982, l.

Flanigen, E.M., "Zeolites: Science and Technology,"
Riberio, F.R.: Rodrigues, A.E.: Rollman, L.D.:
Naccache, C., eds., Martinus Nijhoff, Boston, 1984, 3.

Thomas, J.M.: Williams, C., "Chemical Reactions in
Organic and Inorganic Constrained Systems," D. Reidel,
New York, 1986, 49.

Dwyer, J., Chem. Ind., 1984, 258.

Barrer, R.M., "Zeolites and Clay Minerals as Sorbents
and Molecular sieves,” Academic Press, New York, 1978.

 

Weisz, P.B.: Frilette, V.J., g_. Phys. Chem., 1960, gt,
382.

Csicsery, S.M., Zeolites, 1984, g, 202.
Maxwell, I.E., gt Incl. Phenom., 1986, g, 1.
Derouane, E.G.: Gabelica, 2., g; Catal., 1980, §§, 486.

Corma, A.: Wojciechowski, B.W., Catal. Rev. Sci. Eng.,
1985, 27(1), 29.

Gates, B.C.: Katzer, J.R.: Schuit, G.C.A., "The
Chemistry of Catalytic Processes," McGraw-Hill, New
York, 1974.

 

Breck, D.W., "Zeolite Molecular sieves,” Wiley, New
York, 1974.

190

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

191

Sudo, T.: Shimoda, 8.: Yotsumoto, H.: Aita, 8.,
"Developments in Sedimentology 31: Electron Micrographs
of Clay Minerals," Elsevier, New York, 1981.

Borchardt, G.A., "Minerals in Soil Environments,"
Dixon, J.B.: Weed, J.B., eds., Soil Science Society of
America, 1982, 293.

Pinnavaia, T.J. Science, 1983, 220, 3651.

Scurrell, M.S., "Specialist Periodical Reports.
Catalysis": Royal Society of Chemistry, London, 1978,
2, 215.

Bailar, J.C. Jr., Catal. Rev. Sci. Eng., 1974, l0(1),
17.

Bailey, D. C.: Langer, S. H., Chem. Rev., 1981, 81(2),
109.

Pinnavaia, T.J: Raythatha, R.: Lee, J.G.S., Halloran,
L.J; Hoffman, J., g; Am. Chem. Soc., 1979, 101, 6891.

Mortland, M.M.: Raman, K.V., Clays Clay Miner., 1968,

;§, 393.

Fripiat, J.J., ACS Symp. Ser., 1976, 35, 261.

Mortland, M.M., Trans. 9th Int. Congr. Soil Sci.,
1968, l, 691.

Mortland, M.M.: Fripiat, J.J.: Chaussidon, J.:
Uytterhoeven, J.: L Phys. Chem., 1963, 6_7, 248.

Laszlo, P., Acc. Chem. Res., 1986, $2, 121.

Benesi, H.A.; Winquest, B.H., Adv. Catal., 1978, g1,
170.

Ballantine, J.A., "Chemical Reactions of Organic and
Inorganic Contstrained Systems," Setton, R., ed. , D.
Reidel, Boston, 1986, 197.

Cornelis, A.: Laszlo, P., ibid., 213.

Thomas, J.M., "Intercalation Chemistry," Whittingham,
M.S.: Jacobson, A.J., eds., Academic Press, New York,
1982, 55.

Adams, J.M.: Clapp, T.V.: Clement, D.E.: Reid, P.I.,
g; Mol. Cat., 1984, 31, 179.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

192

Ballantine, J.A.: Davies, 11.: Patel, 1.: Purnell, H.:
Rayanakorn, M.; Williams, K.J.: Thomas, J.M., .J_. _M_ol.
Cat., 1984, gt, 37.

Weiss, A., Angew. Chem. Int. Ed. Engl., 1981, _2_0, 850.

Barrer, R.M: MacLeod, D.M., Trans. Faraday Soc., 1955,
gt, 1290.

Barrer, R.M.: Reay, J.S.S., Trans. Faraday Soc.,
1957, 53, 1253.

Mortland, M.M.: Berkheiser, V.E., Clays Clay Miner.,
1976, g5, 60.

 

Shabtai, 3.: Frydman, N.: Lazar, R., Clays Clay
Miner., 1978, gt, 107.

 

Berkheiser, V.E.: Mortland, M.M., Clays Clay Miner.,
1977, gg, 105.

Traynor, M.F.: Mortland, M.M.: Pinnavaia, T.J., Clayg
Clay Miner., 1978, gt, 313.

' Koppelman, M.H.: Dillard, J.G., Clays Clay Miner.,

1980, gt, 211.

Brindley, G.W.: Sempels, R.E., Clay Miner., 1977, it,
229.

Yamanaka, 8.: Brindley, G.W., Clays Clay Miner., 1978,
.29: 21.

Lahav, N.; Shani, V.: Shabtai, J., Clays Clay Miner.,
1978, gt, 107.

 

Vaughan, D.E.W.: Lussier, R.J., preprints, Fifth
International Conference on Zeolites, Naples, Italy,
June 2-6, 1980.

Shabtai, J.: Lahav, N., U.S. patent 4,216,188 (1980).
Shabtai, J., U.S. patent 4,238,364 (1980).

Lussier, R.J.: Magee, J.S.: Vaughan, D.E.W.,
preprints, Seventh Canadian Symposium on Catalysis,
Edmonton, Alberta, 19-22 October, 1980.

Vaughan, D.E.W.: Lussier, R.J.: Magee, J.S., U.S.
patents 4,176,090 (1979); 4,271,043 (1981): 4,248,739
(1981).

Shabtai, J.: Lazar, R.: Oblad, A.G., New Horiz.
Catal., 1981, l, 828.

 

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

193

Occelli, M.L.: Tindwa, R.M.: Clays Clay Miner., 1983,
3;, 22.

Gaaf, J.: van Santen, R.: Knoester, A.: Winderden, B.,
1; Chem. Soc. Chem. Comm., 1983, 655.

Kikuchi, E.: Hamana, R.: Nakano, M.; Takehara, M.:
Morita, Y., 1; Jpn. Pet. Inst., 1983, gt, 116.

Burch, R.: Warburton, C.I., g_._ Catal., 1986, 21, 503.
ibid., 510.

Endo, T.: Mortland, M.M.: Pinnavaia, T.J., Clays Clay
Miner., 1980, gg, 105.

 

Yamanaka, 8.: Yamashita, G.: Hattori, M., Clays Clay
Miner., 1980, gg, 281.

 

Brindley, G.W.: Yamanaka, S., Amer. Miner., 1979, g,
830.

 

Tzou, M.S., Ph.D. Dissertation, Michigan State
University, 1983.

Yamanaka, 8.: Doi, T.: Sako, S.: Hattori, M., M_at.
Res. Bull., 1984, $2, 161.

Pinnavaia, T.J., Chemical Synthesis using Supported
Reagents,” Laszlo, P., ed., 1986.

Pinnavaia, T.J., "Chemical Reactions in Constrained
Systems," Setton, R., ed., D. Reidel, New York, 1986,
151.

Barrer, R.M., J_. Incl. Phenom., 1986, g, 109.

Lewis, R.M.: Ott, K.C.; van Santen, R.A., U.S. patent
4,510,257 (1985).

Sabatier, P.; Senderens, J.B.: J; Soc. Chem. Ind.,
1902, g;, 504.

 

Badische Anilin und Soda Fabrik, German Patents
293,787 (1913), 295,202 (1914).

Fischer, F.: Tropsch, H., Brennst. Chem., 1923, g,
276.

ibid., 1926, Z, 97.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

194

Masters, C., "Advances in Organometallic Chemistry,"
Stone, F.G.A.: West, R., eds., Academic Press, London,
1964, 63.

Pichler, H., Adv. Catal., 1952, g, 271.

Storch, H.H.; Columbic, N.: Anderson, R.B., "The
Fischer-Tropsch and Related Synthesis," J. Wiley, New
York, 1956, Volume 4.

Anderson, R.B., "Catalysis" Emmett, P.H., ed.,
Reinhold, New York, 1956.

Vannice, M.A., _J_. Catal., 1975, g, 449.

Mills, G.A.: Steffgen, F.W., Catal. Rev., 1974, g,
159.

 

Bell, A.T., Catal. Rev.-Sci. Eng., 1981, 23(1), 203.

Boudart, M.: MacDonald, M.A., J_._ Phys. Chem., 1984,
g, 2185.

 

Ponec, V., "Specialist Periodical Reports. Catalysis":
Royal Society of Chemistry, London, 1982, g, 48.

Vannice, M.A., "Catalysis. Science and Technology,"
Anderson, J.R.: Boudart, M., eds., Springer-Verlag,
Heidelberg, 1982, g, 139.

Biloen, P.: Sactler, W.M.H., Adv. Catal., 1981, 3g,
165.

 

Kellner, C.S.: Bell, A.T., gt Catal., 1981, 19, 418.

Pichler, H.: Schulz, H., Chem.-Ing.-Tech., 1970, gt,
1162.

Dry, M.E., Ind. Eng. Chem. Prod. Res. Dev., 1976, lg,
282.

Henrichi-Olive, G.: Olive, 5., Angw. Chem. Int. Engl.
Ed., 1976, 1;, 136.

Flory, P.J., gt Am. Chem. Soc., 1936, gg, 1877.
Schulz, G.V., g; Phys. Chem., 1935, 39, 399.

King, D.L.: Cusumano, J.A.: Garten, R.L., Catal. Rev.-
Sci. Eng., 1981, 23(1), 233.

Fraenkel, D.: Gates, B.C., gt Am. Chem. Soc., 1979,
102, 2478.

 

87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

98.

99.

100.

101.

102.

103.

104.

105.

106.

195

Nijs, H.H.: Jacobs, P.A.: Uytterhoeven, J.B., L $119.2;
Soc. Chem. Commun., 1979, 180.

 

ibid., 1979, 1095.

Nazar, L.F.: Ozin, G.A.: Hughes, F.: Godber, J.;
Rancourt, D., gt Mol. Cat., 1983, gt, 313.

 

Ballivet-Tkatchenko, D.; Tkatchenko, I., J_. Mol. Cat.,
1981, g;, 1. .

 

Vanhove, D.: Makambo, P.,“ Blanchard, M., J.C.S. Chem.
Comm., 1979, 605.

 

Madon, R.J., g; Catal., 1979, g1, 183.

Bailey, D.C.; Langer, S.H., Chem. Rev., 1981, _8_;, 109.

 

Zwart, J.; Snel, R., gt Mol. Cat., 1985, 39, 305.
Nijs, H.H.: Jacobs, P.A., E; Catal., 1980, §§, 328.
Kellner, C.S.: Bell, A.T., gt Catal., 1982, lg, 251.

Anderson, J.R., "Structure of Metallic Catalysts,"
Academic Press, London, 1975, 369 and 245.

Basset, J.M.; Ugo, R., "Aspects of Homogeneous
Catalysis," Ugo, R., ed., D. Reidel, Boston, 1977,
Volume 3.

Gallezot, P., Catal. Rev. Sci.-Eng., 1979, gg, 21.

Chang, C.D.: Lang, W.H.; Silvestri, A.J., gt Catal.,
1979, §§, 268.

Caesar, P.D.; Brennan, J.A.: Garwood, W.E.: Ciric, J.,
g; Catal., 1979, §§, 274.

Chang, C.D.: Lang, W.H.: Silvestri, A.J., U.S. patent
4,086,262 (1978).

Giannelis, E.P., Ph.D. Dissertation, Michigan State
University, 1984.

Landau, S.D., Ph.D. Dissertation, Michigan State
University, 1984.

McIlwrick, C.R.: Phillips, C.S.G., gt Phys. _E_., 1983,
g, 1973.

Anderson, R.B., ”The Fischer-Tropsch Synthesis,"
Academic Press, New York, 1984.

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

120.

121.

122.

123.

124.

125.

196

Brenner, A.: Hucul, D.A., Inorg. Chem_., 1979, Lg,
2836.

Hughes, F.; Bussiere, P.: Basset, J.M.: Commerenc,
D.: Chauvin, Y.: Bonneviot, L.; Oliver, D., Proc.
Seventh Int. Cong. Catal., Seiyama, T.: Tanabe, R.,
eds., Elsevier, Amsterdam, 1981, A. 418.

Lung, H-J.: Vannice, M.A.: Mulay, L.N.: Stanfield,
R.M.: Delgass, W.N., g_. Catal., 1982, gt, 708.

Sylva, R.N., Pure. gppl. Chem., 1972, g, 115.

Flynn, C.M. Jr., Chem. Rev., 1984, g1, 31.

Arnek, R.: Schlyter, H., Acta. Chem. Scand., 1968,
gg, 1327.

Spiro, T.G.: Allerton, S.E.: Renner, J.; Terzis, A.:
Bliss, R.: Saltman, P., gtAm. Chem. Soc., 1966, g,
2721.

Murphy, D.J.: Posner, A.M.: Quirk, J.P., 1°. Colloig
Interface Sci., 1976, gg, 270.

ibid., 284.
ibid. , 298.
ibid. , 312.

Matijevic, E.; Scheiner, P., g_._ Colloid Interface
Sci., 1978, fig, 509.

Barnhisel, R.I., "Minerals in Soil Environments,"
Dixon, J.B.: Weed, S.B., eds., 8011 Science Society
of America, Madison, WI, 1977, 345.

Oades, J.M., Clays Clay Miner., 1984, gg, 49.

Krebs, H.J.: Bonzel, H.P.: Schwarting, W.: Gafner,
G., g; Catal., 1981, 1;, 199.

Madon, R.J.: Taylor, W.F., E; Catal., 1981, g2, 32.

Dwyer, D.J.: Somorjai, G.A., J_. Catal., 1978, g,
291.

Dwyer, D.J.: Hurdenbergh, H., J_. Catal., 1984, g1,
66, and references therein.

Raupp, G.B.; Delgass, W.N., g; Catal, 1979, gg, 361.

126.

127.

128.

129.

130.

131.

132.

133.

134.

135.

136.

137.

138.

139.

140.

141.

142.

143.

197

Niemantsverdriet, J.W.: van der Kraan, A.M.: van der
Dijk, W.L,: van der Baan, 11.8., :1; Phys. Chem., 1980,
gt, 3363.

 

Dictor, R.A.: Bell, A.T., J_. Catal., 1986, g1, 121.

Accuri, K.B.: Schwartz, L.H.: Piotrowski, R.D.: Butt,
J.B., g; Catal., 1984, 2;, 349.

Amelse, J.A.: Schwartz, L.H.: Butt, J.B., _J_. Catal._.,
1981, 1g, 95.

Pichler, H.: Schulz, H., Chem. Ingt Tech., 1970, _4_2,
1162.

Konig, L.; Gaube, J., Chem. Ing. Tech., 1983, gg, 14.

Huff, G.A. Jr.: Satterfield, C.N., g; Catal., 1984,
gg, 37o.

Gard, J.S., ed., "The Electron-Optical Investigation
of Clays," Mineralogical Society, London, 1971.

2vyagin, B.B., ”Electron Diffraction Analysis of Clay
Mineral Structures," Plenum Press, New York, 1967.

Thomas, J.M.; Jefferson, D.A.: Mullinson, L.G.:
Smith, D.J.: Crawford, E.S., Chem. Scripta., 1978-
79, it, 167.

Cowley, J.M.: Goswomi, A., Acta. Cryst., 1961, _J_.g,
1071.

Goldstein, J.I., "Practical Scanning Electron
Microscopy," Goldstein, J.I.: Yakowitz, H., eds.,
Plenum Press, New York, 1975.

Cliff, G.: Lorimer, G.N., J_.Microsc., 1975, 103,
203.

Pinnavaia, T.J.: Tzou. M.S.: Landau, S.D.: Raythatha,
R.H., g; M01. Cat., 1984, g1, 195.

Occelli, M.L.: Tindwa, R.M., Clays Clay Miner., 1983,
31( ), 22.

Occelli, M.L.: Landau, S.D.: Pinnavaia, T.J., _J_.
Catal., 1984, 29, 256.

Occelli, M.L.: Landau, S.D.: Pinnavaia, T.J.,
submitted to E; Catal., 1986.

van Damme, H.: Fripiat, J.J., 2:. Chem. Phys., 1985,
gg, 2785.

144.

145.

146.

147.

148.

149.

150.

151.

152.

153.

154.

155.

156.

157.

158.

159.

160.

161.

162.

163.

198

Pinnavaia, T.J., personal communication.

Plee, D.: Borg, F.: Gatineau, L.: Fripiat, J.J., _J_._
Am. Chem. Soc., 1985, 107, 2363.

Schmidt, L.D.: Wang, T.: Vacques, A.,
Ultramicroscopy, 1982, g, 175.

Delannay, F3, ”Characterization of Heterogeneous
Catalysts," Marcel Dekker, New York, 1984, 71.

Lyman, C.E., g; M61. Cat., 1983, g9, 357.

Cowley, J.M., "Catalytic Materials: Relationship
Between Structure and Reactivity," Whyte, TLE. Jrz:
Della Betta, R.A.: Derovane, E.G.: Baker, R.T.K.,
American Chemical Society, Washington, DAL, 1984,
353.

Kuznetzov, V.L.: Bell, A.T.: Yermakov, Y.I., _J_._
Catal., 1980, gg, 374.

Chen, Y.W.: Wang, H.T.; Goodwin, J.G. Jr., Shiflett,
W.K., Appl. Catal., 1983, g, 303.

 

Egiebor, N.O., 91 M01. Chem., 1985, g, 253.

Dalla Betta, R.A.: Shelef, M., _J_._ Catal., 1977,
fig, 111.

King, D.L., gt Catal., 1980, fig, 77.

Kellner, C.S.: Bell, A.T., J: Catal., 1981, 1;, 296.

Guglielminotti, E.: Zecchina, A.: 80831, A.: Camia,
M., gg_Catal., 1982, 11, 225.

1111,, 240.

gptgg, 252.

Rightor, E.G.: Pinnavaia, T.J., to be published,

Ultramicroscopy, 1987.

Flynn, P.C.: Wanke, S.E., {L Catal., 1974, g, 396.

Pulvermacher, B.: Ruckenstein, B., gt Catal., 1974,
3;, 115.

Avery, N.R.: Sanders, J.V., g; Catal., 1970, gg, 129.

Volta, J.G.; Tatibouet, J.M., J_. Catal., 1985, 25,
467.

164.

165.

166.

167.

168.

169.

170.

171.

172.

173.

174.

175.

176.

177.

178.

179.

180.

181.

199

Lee, H.H.; Ruckenstein, E., Catal. Rev.-Sci. Eng.,
1983, 25(4), 475.

 

King, D.L., g; Catal., 1978, g), 386.

Okuhara, T.: Kumura, T.: Kobayashi, H.: Misono, M.;
Yoneda, Y., Bull. Chem. Soc. Japan, 1984, g1, 938.

Goodwin, J.G. Jr.: Goa, D.O.: Erdal, S.: Rogan, F.H.,
Appl. Catal., 1986, gt, 199.

Treacy, M.M.J.: Howie, A., g_._ Catal., 1980, g, 265.

Crewe, A.V.: Langmore, J.P.: Isaacson, M.S.,
"Physical Aspects of Electron Microscopy and
Microbeam Analysis," Siegel, B.M.: Beaman, D.R., J.
Wiley, New York, 1975, 47.

Cowley, J.M., "Scanning Electron Microscopy/1982,"
Johari, ed., (SEM, AMF, O'Hare, IL, 1983), volume 1,
51.

Howie, A.: Marks, L.D.: Penneycook, S.J.,
Ultramicroscopy, 1982, g, 175.

Schulz, H., 91 M01. Chem., 1985, g, 231.

Kellner, s.s.; Bell, A.T., g; Catal., 1981, 19, 418.

Chen,

Y.W.: Wang, H.T.: Goodwin, J.G. Jr., g_._ Catagg,
1984,

gg, 499.

Chen, Y.W.: Wang, H.T.: Goodwin, J.G. Jr., _J_. Catal.,
gg, 415.

Tatsumi, T.: Shul, Y.G.: Sugivra, T.: Tominaga, H.,
Appl. Catal,, 1986, gt, 119.

Okuhara, T.: Tamura, H.: Misono, M., J_. Catal., 1985,
3;, 41.

Wang, H.T.: Chen, Y.W.: Goodwin, J.G. Jr., Zeolites,
1984, g, 57.

 

Winslow, P.: Bell, A.T., J_. Catal., 1985, a: 142.

Egiebor, N.O.: Cooper, W.C., Appl. Catal., 1985, H,
47.

Chuang, S.C.: Tian, Y.H.: Goodwin, J.G. Jr.: Wender,
1., Q; Catal., 1985, gg, 396.

200

182. Sajkowski, D.J.: Lee, S.Y.: Schwank, J.: Tian, Y.H.:
Goodwin, J.G., g; Catal., 1986, 21, 549.

183. Lineweaver, L.J., gt gppl. Phys., 1983, gt, 1786.

 

184. Thomas, L.E., Ultramicroscqpy, 1985, gt, 173.
185. Crewe, A.V., Chem. Scripta., 1978-79, )5, 17.
186. Harris, W.J., Nature, 1962, 196, 499.

187. Brunauer, S.: Emmett, P.H.: Teller, E., _J_._Am. Chem.

 

Soc., 1938, g9, 5.

188. Lowell, S.: Shields, J.B., Powder Surface Area and
Porosity, Chapman and Hall, New York, Chapter 4.

189. Buessemeier, B.: Frohning, C.D.: Cornils, B.,
Hydrocarbon Processing, November 1976, 55(3), 105.