Citation
Further developments and application of the method of selective physisorption for measuring active catalyst surface area

Material Information

Title:
Further developments and application of the method of selective physisorption for measuring active catalyst surface area
Creator:
Toor, Irfan Ali ( Dissertant )
Lee, Hong H. ( Thesis advisor )
Hoflund, Gar B. ( Reviewer )
Westermann-Clark, Gerald B. ( Reviewer )
O'Connell, John P. ( Reviewer )
Allen, Eric R. ( Reviewer )
Place of Publication:
Gainesville, Fla.
Publisher:
University of Florida
Publication Date:
Copyright Date:
1985
Language:
English
Physical Description:
vii, 116 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Betting ( jstor )
Calibration ( jstor )
Carbon ( jstor )
Catalysts ( jstor )
Chemisorption ( jstor )
Desorption ( jstor )
Nitrous oxide ( jstor )
Oxides ( jstor )
Platinum ( jstor )
Surface areas ( jstor )
Catalysts -- Measurement ( lcsh )
Chemical Engineering thesis Ph. D
Dissertations, Academic -- Chemical Engineering -- UF
Gases -- Absorption and adsorption ( lcsh )
Metal catalysts ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
Modifications have been made in the method of selective physisorption as developed by Miller and Lee (Miller, D.J., and H.H. Lee, AIChE J., 30, 84 (1984)). The new method makes use of packing factors of the selectively physisorbed gas instead of fractional coverage. The spread of packing factors versus temperature curves has been shown to be much greater than the spread of fractional coverage versus temperature curves. The method has been tested on a standard catalyst system of platinum supported on silica. The results have been compared with the results of hydrogen chemisorption and it has been observed that they are in fair agreement. The application of the new method has been extended to metal oxide catalysts. The results indicate that nitrous oxide is not the best suited adsorbate for the moly-alumina catalyst, in the sense that the nitrous oxide selective physisorption does not yield the absolute value of the fractional surface area of this catalyst. However, it can still be used to determine the ratios of the fractional surface areas of different molyalumina catalysts with up to 15 wtXMoO^ loading. The values of the fractional surface area ratios calculated from the nitrous oxide selective physisorp ti on results were much lower than the values computed from the BET results, assuming an epitaxial monolayer of molyoxide on the surface of alumina. It was shown that the cyclohexane dehydrogenation activity increases with the increase in nitrous oxide packing factors for the moly-alumina catalysts. The nitrous oxide selective physisorption was not successful on moly-silica catalyst.
Thesis:
Thesis (Ph. D.)--University of Florida, 1985.
Bibliography:
Bibliography: leaves 114-115.
Additional Physical Form:
Also available on World Wide Web
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Irfan Ali Toor.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
020859655 ( AlephBibNum )
12647182 ( OCLC )
ACU5626 ( NOTIS )

Downloads

This item has the following downloads:


Full Text











FURTHER DEVELOPMENTS AND APPLICATION OF THE METHOD OF
SELECTIVE PHYSISORPTION
FOR MEASURING ACTIVE CATALYST SURFACE AREA








By

Irfan Ali Toor


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


UNIVERSITY OF FLORIDA


1985













ACKNOWLEDGMENTS

The author wishes to express his profound gratitude

towards Dr. Hong H. Lee, chairman of the advisory commit-

tee, for suggesting this project and for his patience and

continued guidance throughout this work.

The author also wishes to acknowledge his gratitude

towards Dr. Gar B. Hoflund, Dr. J.P. O'Connell, Dr. G.B.

Westerman-Clark and Dr. E.R. Allen for serving on the

advisory committee and for their valuable criticism ot the

manuscript. Sincere appreciation is also extended to Mr.

Tracy Lambert, Mr. Ron Baxley and Mr. Rudi Strohschein for

their skilled services in fabricating the experimental

apparatus.

Finally, the author wishes to thank his parents, broth-

ers, sisters, wife and child for their patience and invalu-

able moral support without which the completion of this work

would not have been possible.













TABLE OF CONTENTS



Page

ACKNOWLEDGMENTS...................................... ii

KEY TO SYMBOLS........................................ v

ABSTRACT............................................ vi

CHAPTERS

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

II MODIFICATIONS AND IMPROVEMENTS IN THE
METHOD OF SELECTIVE PHYSISORPTION .............. 7

Experimental Modifications....................... 8
Theoretical Modifications........................ 8

III APPLICATION OF THE SELECTIVE PHYSISORP-
TION METHOD TO METAL CATALYST.................. 15

Experimental Apparatus And Procedure............ 16
Catalyst Preparation............................. 18
Results And Discussion........................... 19

IV APPLICATION OF THE METHOD OF SELECTIVE
PHYSISORPTION TO OXIDE CATALYST................ 31

Experimental Apparatus........................... 32
Experimental Procedure........................... 38
Catalyst Preparation ........................... 40
Experimental Results And Discussion............ 41
Summary............. .............................. 68

V CONCLUSIONS AND RECOMMENDATIONS................ 73

APPENDICES

A PREPARATION OF OXIDE CATALYSTS................. 80









B NITROUS OXIDE SELECTIVE PHYSISORPTION DATA FOR
PLATINUM SILICA CATALYST......................... 82

C NITROUS OXIDE SELECTIVE PHYSISORPTION DATA FOR
THE OXIDE CATALYSTS............................ 86

D NITROGEN PHYSISORPTION RAW DATA................ 90

E DIFFERENTIAL REACTOR DATA........................ 110

F ERROR ANALYSIS OF THE PACKING FACTOR VALUES.... 113

LIST OF REFERENCES................................... 114

BIOGRAPHICAL SKETCH .................................. 116












KEY TO SYMBOLS
b. proportionality constant in Eq. (10)

11 integral defined by Eq. (7) or by Eq. (8)

It integral defined by Eq. (8) or by Eq. (12)

R1 volume of gas adsorbed on catalyst per unit surface
area of the catalyst in the supported state

R2 volume of gas adsorbed on support per unit area of the
support in the supported state

S1 catalyst surface area
S2 support surface area

St total surface area (S1+S2)
T temperature

Ti lower limit for integration with respect to temperature

Tf upper limit for integration with respect to temperature

Ui volume of gas adsorbed on pure catalyst (i=l) or pure
support (i=2)
v1 volume of gas adsorbed on the catalyst in the supported
state

v2 volume of gas adsorbed on the support in the supported
state

vt V1 +V2

Y1 U1/S1

Y2 U2/S2

Yt Vt/St
Subscript i = 1 for catalyst, 2 for support













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


FURTHER DEVELOPMENTS AND APPLICATION OF THE METHOD OF
SELECTIVE PHYSISORPTION
FOR MEASURING ACTIVE CATALYST SURFACE AREA


By

Irfan Ali Toor

May, 1985

Chairman: Dr. H.H. Lee
Major Department: Chemical Engineering

Modifications have been made in the method of selective

physisorption as developed by Miller and Lee (Miller, D.J.,

and H.H. Lee, AIChE J., 30, 84 (1984)). The new method makes

use of packing factors of the selectively physisorbed gas

instead of fractional coverage. The spread of packing factors

versus temperature curves has been shown to be much greater

than the spread of fractional coverage versus temperature

curves. The method has been tested on a standard catalyst sys-

tem of platinum supported on silica. The results have been

compared with the results of hydrogen chemisorption and it has

been observed that they are in fair agreement. The applica-

tion of the new method has been extended to metal oxide

catalysts. The results indicate that nitrous oxide is not the

best suited adsorbate for the moly-alumina catalyst, in the








sense that the nitrous oxide selective physisorption does not

yield the absolute value of the fractional surface area of

this catalyst. However, it can still be used to determine the

ratios of the fractional surface areas of different moly-

alumina catalysts with up to 15 wt%MoO3 loading. The values of

the fractional surface area ratios calculated from the nitrous

oxide selective physisorption results were much lower than the

values computed from the BET results, assuming an epitaxial

monolayer of moly oxide on the surface of alumina. It was

shown that the cyclohexane dehydrogenation activity increases

with the increase in nitrous oxide packing factors for the

moly-alumina catalysts. The nitrous oxide selective physisorp-

tion was not successful on moly-silica catalyst.













CHAPTER 1
INTRODUCTION

Catalytic materials are prepared with the intention of

maximizing the exposed surface area of the active material

upon which the reaction may take place. Many important

industrial catalysts are prepared by dispersing the active

material on the surface of a porous material generally

referred to as the support. Since the number of the

catalytically active sites is believed to be proportional to

the total surface area of the dispersed material it is

desirable to measure the surface area of the dispersed

catalyst. Unfortunately, direct measurement of the surface

area of the catalyst material dispersed in the pores of the

support is not possible at the present time. Therefore vari-

ous indirect techniques have been developed to characterize

the dispersed catalyst. X-ray line broadening (Anderson ,

1968), mossbaur spectroscopy (Bartholomew and Boudart,

1973), magnetization techniques (Hill and Selwood, 1949),

and low angle scattering (Sinfelt, 1967), to mention just a

few. Another important method of characterizing the porous

catalysts is the physical and chemical adsorption of dif-

ferent gases to determine the surface area of the dispersed

catalyst. Brunauer, Emmett, and Teller (1938) were the first

to use the low temperature physical adsorption isotherms of









various gases to measure the total surface area of the

porous catalysts. This method, commonly known as the BET

method, has found wide application in catalytic research all

around the world. Brunauer and Emmett (1938, 1940) also stu-

died the selective chemisorption of various gases to deter-

mine the surface area of dispersed catalysts. Since that

time, numerous attempts have been made to use the method of

selective chemisorption to determine the fractional surface

area of oxide catalysts but they have met with limited suc-

cess.

The chemisorption method differs from the physisorption

method in that, ideally, the chemical adsorption occurs only

onto a particular component of the catalytic material. The

method requires that the adsorbate form a monolayer of the

chemisorbed atoms on the surface of the catalyst and that

there exist a simple relationship between the number of

molecules or atoms adsorbed and the number of surface atoms

of the catalyst. Since most metals chemisorb small molecules

like H2, 02, and CO the chemisorption has been quite suc-

cessful in characterizing many metal catalysts. For instance

hydrogen chemisorption has been used very effectively to

determine the fractional surface area of dispersed platinum

catalysts (Spenedal and Boudart, 1960; Adler and Kearney,

1960). In the case of the metal oxide catalysts, however,

the situation is further complicated because of two prob-

lematic points : (a) the lower interaction specificity of

the possible gas/metal oxide system relative to that for








gas/support system, and (b) the necessity to look for dif-

ferent gases for each supported metal oxide catalyst.

Nevertheless, numerous attempts have been made to use selec-

tive chemisorption of gases to characterize metal compound

and metal oxide catalysts (Weller and Voltz, 1954 ; Segawa

and Hall, 1983) but difficulty in the interpretation of the

experimental results still remains. A series of articles has

been published by Weller and associates (Parekh and Weller,

1977; Parekh and Weller, 1978; Srinivasan, Liu, and Weller,

1979; Liu, Yuan, and Weller, 1980; Garcia Fierro, Mendioroz,

Pajares, and Weller, 1980; Liu and Weller, 1980) to discuss

the low temperature oxygen chemisorption for measuring the

fractional surface area of molybdenum oxide dispersed on

alumina but their results are at variance with current ideas

concerning the formation of an epitaxial monolayer of

molybdenum oxide on the surface of alumina. It has also been

shown that in oxide catalysts the chemisorption is site

selective (Millman, Crespin, Crillo, Abdo and Hall, 1979;

Vaylon and Hall, 1983) and therefore represents only a frac-

tion of the actual surface area of the dispersed catalyst.

From the above discussion it is clear that there was a need

to look into new methods of gas adsorption of porous

catalysts which may be applicable to a wider range of

catalysts including metal and metal oxide catalysts.

A new experimental method of measuring the active sur-

face area of supported catalysts using selective physisorp-

tion of certain gasses has been put forward recently by








Miller and Lee (1984). The basic idea here was that a mono-

or sub-monolayer volume of a gas physisorbed on a two com-

ponent solid as in a supported catalyst should be able to

distinguish between the two different surfaces upon thermal

desorption if the interactions between the gas and the sur-

face are strong, say 3-6 Kcal/mole in terms of heats of

adsorption. The desorption characteristics of gas coverage

with temperature for the catalyst, the support, and the sup-

ported catalyst were used by Miller and Lee (1984) to calcu-

late the fraction of the total surface area occupied by the

catalyst. Because of the nature of physisorption the selec-

tive physisorption should be applicable to any supported

catalyst including oxides and metal compound catalysts pro-

vided a suitable adsorbate is used. This is in contrast to

the chemisorption method (Spenedal and Boudart, 1960; Adler

and Kearney, 1960) which does not yield the catalyst surface

area of some metal catalysts and the majority of metal com-

pound catalysts because of the specific nature of chemisorp-

tion, although it yields valuable information of the sites

active to a chemisorbing gas.

In the original development (Miller and Lee, 1984),

thermal desorption of selectively physisorbed carbon dioxide

on the surface of potassium carbonate carbonate-carbon black

mixtures was used to determine the fractional surface area

of each component in the physical mixture. A qualitative

comparison between the selective physisorption method and

oxygen chemisorption was also given for supported silver




5


catalyst. Nevertheless, no conclusive evidence was presented

as to the effectiveness of the selective physisorption

method as applied to the supported catalysts. Further the

thermal desorption experiments and the data analysis were

such that the method was effective only when catalyst load-

ings were high and the catalyst covered a significant por-

tion of the total surface area of the support. In Chapter 2,

we present refinements in the original method that would

allow a more accurate determination of the catalyst surface

area even when the catalyst covers a small portion of the

support surface area. The new method involves the use of

packing factors (y) in place of fractional gas coverage (e)

to calculate the fractional surface area of the dispersed

catalyst.

The effectiveness of the selective physisorption method

in measuring fractional catalyst surface area of dispersed

catalyst is demonstrated in the Chapter 3. A catalyst system

was sought that is amenable to a well established method of

measuring the fractional surface area of the dispersed

catalyst. Platinum dispersed on silica was chosen because

hydrogen chemisorption has been well established as a reli-

able method measuring the surface area of dispersed platinum

and thus a valuable comparison of results could be made with

the results of selective physisorption.

Since an important advantage of the method of selective

physisorption would be a more general and wider application

to different types of catalysts, it was decided to test the




6



new method on at least one industrially important metal

oxide catalyst. The application of the selective physisorp-

tion method to moly-alumina catalyst is presented in Chapter

4. The results are correlated with cyclohexane dehydrogena-

tion activity of the catalyst measured in a differential

reactor. Conclusions and recommendations for future work are

discussed in the Chapter 5.













CHAPTER 2
MODIFICATIONS AND IMPROVEMENTS IN THE METHOD OF
SELECTIVE PHYSISORPTION

Selective physisorption utilizes the differences in the

forces of interaction between different gas solid pairs. A

detailed discussion of the nature of the physisorption

forces and the kinetic models used to describe this

phenomenon has been given by Miller (1982) and will not be

presented here. Anyone interested in a more thorough under-

standing of the subject is referred to Brunauer (1943),

Young and Crowell (1962), Clark (1970), Ricca (1972), Herz

et al. (1982). The purpose of this research is to advance

the theory of selective physisorption as applied to the

measurement of the fractional surface area of supported

catalyst. A few modifications have been made in the experi-

mental method and the method of data analysis which render

the methods of selective physisorption more sensitive to

changes in the fractional surface area of the dispersed

catalyst. These modifications and improvements will be dis-

cussed in the following paragraphs. The main contribution

of this study, however, is the demonstration of the effec-

tiveness of the selective physisorption in determining the

fractional surface area of supported metal catalysts

(Chapter 3), and its extension to oxide catalysts (Chapter

4).








Experimental Modifications

In the original development, Miller and Lee (1984)

utilized the differences in the gas coverage (6) versus tem-

perature (T) relationships of the mono- or sub-monolayer

volume of a physisorbed gas to determine the fraction of the

total surface area covered by the catalyst. In order to gen-

erate the thermal desorption isobars it was assumed that a

pseudo steady state exists during the transient thermal

desorption of the mono- or sub-monolayer of the physisorbed

gas. Therefore a single thermal desorption was sufficient to

obtain a 6-T relationship (Miller and Lee, 1984). The pseudo

steady state assumption was avoided in this study and

instead more than one constant temperature cold baths were

used to determine the amount of gas adsorbed at each tem-

perature in order to generate a more accurate G-T relation-

ship. The motivation is that the total desorption will pro-

vide a more accurate determination of the amount of gas

adsorbed and thus a more accurate adsorption isobar will be

achieved which in turn will enhance the accuracy with which

the catalyst surface area is determined.

Theoretical Modifications

While Miller and Lee (1984) have used the 8-T relation-

ship to determine the fractional surface area of the sup-

ported catalyst, it was found that the use of packing factor

(which is the amount of gas adsorbed per unit surface area

of the supported catalyst) in place of e would allow a more

accurate determination of the fractional surface area of the







supported catalyst. For instance, consider Figure 1, which

compares the 6-T curves of the pure components of potassium

carbonate and carbon black mixture, obtained by Miller and

Lee (1984), with the spread of the Y (the packing factor

expressed as volume of gas per unit N2 BET area) versus T

curves generated from the same data. It is clear that the

spread of the y-T curves is much greater than the spread of

the e-T curves and therefore the former relationship will

provide a better sensitivity towards the determination of

the fractional surface area of the dispersed catalyst. In

light of the above development it was necessary to develop

an expression for the determination of the fractional sur-

face based on the packing factors rather than 6 in order to

take advantage of the greater sensitivity afforded by the

packing factor.

If we let v1 and v2 be the volumes of the submonolayer

of a gas adsorbed on the surfaces 1 (catalyst) and 2 (sup-

port) in the supported catalyst then the total volume of the

gas adsorbed on the supported catalyst vt is simply the sum

of v1 and v2:



vt(T) = v1(T) + v2(T). (1)


Further if we let x.(T) be the fraction of the surface area

of the surface i that is covered by the adsorbate at the

temperature T, the total surface area of the supported

catalyst, St, is given by




10














.16 .1.4


.14. K2C03 .1.2
S, O
"N Y
.12. 1\ .0

S-



.08 \ 6













-80 -60 -40 -20 0 20
T (C)

figure I. Comparison between y-T curves and 8-T
curves for the potassium carbonate-carbon
black system.
. N, No

.08 0 .6'Z










black system.








V1 1 v2 1
S -+ (2)
t x1 R1 x2 R2 (2)



where the factor R. is given by
1


S cm adsorbate on surface i (
R.= (3)
m surface area of surface i


Since v. cannot be measured directly, but vt can be measured

using thermal desorption, we rearrange Eq. (2) with the help

of Eq. (1) to get


v (T)
t R x
2 2
v1(T) = 1 1 (4)
R1X1 R2x2


Further the fractional surface area of the dispersed

catalyst, S1 / St, is given by


v (T)
1 [ ] [ ]
S 1 [ ] (
S t R 1 R1X (T)
R2x2(T)


which follows from Eq. (4) and the relationship

S1 = v1(T) / R1x1(T). It is important to note that the rela-
tionship of Eq. (5) yields a value of the fractional surface

area of the dispersed catalyst at every temperature at which

physisorption can be obtained. Therefore Eq. (5) can be








integrated over the entire temperature range of interest and

the result will still be valid. This is done in order to

take advantage of the entire adsorption isobar. Also the

integration cancels out the scatter in the experimental data

and the calculated value will be the average value of the

fractional surface area of the dispersed catalyst,

representing the temperature range of interest of the

adsorption isobar. Multiplying SI / St by the numerator of

Eq. (5), integrating, and rearranging yield the following

final result


S1 I
(6)
St t

where



1 = 1 x ] dT (7)
f t R2x
T.
1

and



It T [ 1 ] dT. (8)
t R2x2
T.


Here Ti and Tf are the temperatures chosen for integration

in the temperature range of interest.

The values of R.xi can readily be obtained experimen-

tally for the two pure components constituting the supported








catalyst. The definitions of R. and x., when applied to the

pure components, yield


U.
S x (T) = Yi(T) (9)
1 1 pure ( N BET area ) ipu()
2 B,pure


where U. is the sub-monolayer volume of the gas adsorbed on

pure catalyst (i=1) and pure support (i=2) surfaces at tem-

perature T. It should be noted that U. is different from v..
1 1
For the platinum catalyst supported on silica (Chapter 3),

for instance, the volume of gas adsorbed on the platinum

particles is U. whereas the volume of gas adsorbed on the

platinum particles dispersed in the supported catalyst is

v1. If we assume that the quantity R.xi for the surface

unsupported state is proportional to that for the surfaces

in the pure state, we have



R.x.(T) = b. Yi(T) (10)



where b. is the proportionality constant. However, if the

adsorbate is indeed physisorbed it will not distinguish

between pure and dispersed states of the catalyst and there-

fore the constants b. should assume a value of unity. For
1
such an adsorbate, Rixi is equal to yi and all the quanti-

ties in Eqs. (7) and (8) can readily be obtained from exper-

iments, thus allowing the calculation of the fractional








catalyst surface area from Eq.


(6) because Eqs. (7) and (8)


reduce to


11 f


T.



T.
i'


(1 ) dT
(1 ) dT


'l
(1 - ) dT
Y2


where Yt = v / S.
l- t ~t


(11)


(12)














CHAPTER 3
APPLICATION OF THE SELECTIVE PHYSISORPTION METHOD
TO METAL CATALYSTS

One shortcoming of the work of Miller and Lee (1984)

was that no conclusive evidence was provided as to the

effectiveness of the selective physisorption method as

applied to the dispersed catalysts. They used the selective

physisorption of carbon dioxide at -78 0C to determine the

fractional catalyst surface area of silver catalyst

dispersed on alumina and compared the results with those of

oxygen chemisorption. However, the stoichiometry of oxygen

chemisorption on dispersed silver catalyst is not fully

understood (Miller, 1982) and is not a well established

method of measuring the fractional surface area of silver

catalyst. Furthermore the comparison made by the original

authors was only qualitative in nature, and therefore does

not furnish a positive verification of the surface area cal-

culated using thermal desorption of the selectively phy-

sisorbed carbon dioxide. Scanning electron microscopy was

also attempted to determine the surface area of dispersed

silver but the results were inconclusive.

The other catalyst system studied by Miller and Lee

(1984) was potassium carbonate dispersed on carbon black.

Again no independent measurement of potassium carbonate sur-

face area could be made and thus no comparison was made.








Physical mixtures of carbon black and potassium carbonate

were then used to partially verify the results but such a

comparison can only be of a limited value.

In order to provide a positive evidence that the selec-

tive physisorption method can be used effectively to deter-

mine the fractional surface area of dispersed metal

catalyst, a catalyst system was sought that would provide an

already established alternate method of independently

measuring the fractional surface area of the catalyst so

that the results of the selective physisorption method could

be verified. Platinum supported on silica was chosen because

it meets these conditions perfectly. Hydrogen chemisorption

has been studied extensively (Spenedal and Boudart, 1960;

Adler and Kearney, 1960; Wanke, Lotochinski and Sidwell,

1981; Sarakany and Gonzalez, 1982) and is a well established

method of measuring the fractional surface area of platinum

catalyst. Therefore it was decided to determine the frac-

tional surface area of platinum in platinum-silica catalyst

using the new method and to verify the results by making a

comparison with the platinum surface area determined from

hydrogen chemisorption. In this manner, it would be possible

to demonstrate the effectiveness of the selective physisorp-

tion method in determining the fractional surface area of

metal catalysts.

Experimental Apparatus And Procedure

The gas adsorption experiments in this study have been

carried out with the help of a Perkin Elmer continuous flow








sorptometer modified for thermal desorption experiments. A

detailed description of this apparatus has been given by

Miller (1982) and will not be presented here.

The constant temperature cold baths that were used for

low temperature adsorption were made from liquid nitrogen,

acetone-dry ice, chloroform-dry ice, and chlorobenzene-dry

ice mixtures to obtain -196 0C, -78 0C, -61 0C, and -45 C

(Phipps and Hume, 1968) respectively. The sample cell con-

taining the catalyst sample (Miller, 1982) was immersed in

the cold bath and the system was allowed to equilibrate.

Equilibrium was determined by a constant reading of the tem-

perature recorder for a period of 4-5 minutes, and a stable

response from the thermal conductivity cell detector. After

the equilibrium had been reached the cold bath was removed

instantaneously and the sample was heated with the help of

an external heating coil, to completely desorb the gas. The

response of the thermal conductivity cell was recorded using

an integrator recorder. A minimum of two readings were taken

at each adsorption temperature. A 0.5 ml pulse of the

adsorbate gas was used to calibrate the response of the

thermal conductivity cell.

All samples were processed one after the other within a

period of 24 hrs. to ensure minimal fluctuations in the gas

flow rate. The gas flow rates were measured with a soap bub-

ble meter after the last sample had been processed, as

detailed in the reference.







The hydrogen chemisorption experiments were carried out

using the dynamic pulse method (Wanke, Lotochinski and

Sidwell, 1981). A detailed description of the method and the

experimental apparatus is available in the reference and

need not be presented here. Pulses of 0.1 ml of hydrogen

(Airco Grade 5.5, 99.9995%) were used for the chemisorption

at 100 C. Each gas was purified by first passing through a

bed of 4 A molecular sieve (Davison Chemical, Grade 5.3, 4-8

mesh beads) and then through a bed of copper.

In the BET and the low temperature adsorption experi-

ments, helium (Airco Grade 4.5, 99.995%) was used as the

carrier gas with nitrogen (Linde, Ultra High Purity,

99.999%), carbon dioxide (Airco Grade 4, 99.99%) or nitrous

oxide (Matheson, Ultra High Purity, 99.999%) as the adsor-

bate. Each gas was purified by first passing through a 4A

molecular sieve (Davison Chemical Grade 5.3, 4-8 mesh beads)

which was followed by a cold trap. The cold trap consisted

of liquid nitrogen in a dewer flask when nitrogen was used

as adsorbate, or acetone-dry ice slush in a dewer flask when

nitrous oxide of carbon dioxide was used as the adsorbate.

Catalyst Preparation

Platinum-silica catalysts were prepared by impregnating

powdered silica (Alpha, 99.5% Si02, -400 mesh amorphous)

with chloroplatanic acid solution. A stock solution of

chloroplatanic acid was prepared by dissolving 5.1 gm of

chloroplatanic acid (Alpha) in 100 ml of distilled water.

Impregnation was carried out by adding 11.5, 24.5 ml, and








36.0 ml of the stock solution to three 5 gm samples of sin-

tered silica to prepare 5 wt%, 10 wt%, and 15 wt%Pt

catalysts respectively. Each sample was allowed to stand for

a 24-hr period for impregnation. Intermittent stirring was

continued during this 24-hr period. The samples were then

dried in an oven at 120 0C for 24 hrs, and then reduced in

flowing hydrogen at 500 C for 17 hrs.

Another catalyst sample with 1 wt%Pt loading was

prepared in the following manner. A calculated amount of

chloroplatanic acid which gives 1 wt%Pt loading on 10 grams

of silica was dissolved in 5 grams of distilled water. The

resulting solution was added to 10 grams of amorphous silica

(Alpha, 99.5%Si02, -400 mesh amorphous)to obtain a thick

slurry. This was allowed to stand for a 24-hr period for

impregnation. During this time, intermittent stirring was

continued to maintain a uniform slurry. At the end of the

24-hr period, the impregnated slurry was dried at 120 C for

another 24 hrs. This process was followed by 16 hrs of

reduction at 500 C in a stream of flowing hydrogen. No pre-

treatment was carried out to increase the dispersion of any

catalyst.

Results And Discussion

The BET and hydrogen chemisorption results are summar-

ized in Table 1. The BET total surface area for each sample

was determined using four data points at four different

pressures. A one-to-one hydrogen-to-platinum correspondence

was used to convert the amount of hydrogen chemisorbed on








Table 1. BET and H2 chemisorption results.


N monoleyer Amount of H2 *
volume at STP BEJ Area chemisorbed 3 Active area
Sample (ml/gcat) (m /gcat) (ml H2/gcat *10 ) (m Pt/gcat)


Pt particles 0.16 0.70 - -

Si02 1.53 6.66

1 wt% Pt on Si02 1.22 5.79 67.6 0.324

5 wt% Pt 1.19 5.66 '7.7 0.037

10 wt% Pt 1.11 5.26 59.0 0.283

15 wt% Pt 0.96 4.57 54.7 0.262

*Area calculated from the amount of hydrogen chemisorbed.








the active platinum surface area of the supported catalyst,

using a value of 1.12 x 1019 Pt sites/m2 (Spenedal and

Boudart, 1960). It is interesting to note in Table 1 that

the total BET surface per gram of supported catalyst

decreases with increasing platinum loading. This behavior

could be due to a growing number of platinum crystallites as

the platinum loading is increased. The large crystallites

may block micropores and thereby cause a reduction in the

total surface area of the supported catalyst.

The chemisorption results reported in Table 1

(equivalently in Table 2) need some comments. As indicated

earlier, the impregnation technique for the 1 wt%Pt catalyst

was different from those for 5, 10, 15 wt%Pt samples. As

apparent from the table, inefficient impregnation for the

higher loading catalysts resulted in low platinum disper-

sion. Because of the unusual behavior, the chemisorption

experiments were repeated only to confirm the reported

values in Table 2. This apparent drawback, however, was

instrumental in demonstrating that selective physisorption

can distinguish small differences in fairly small surface

areas which is an important feature because the technique

was shown earlier (Miller and Lee, 1984) to be effective

only for large surface areas. One can see for the higher

loading sample catalysts that the catalyst surface area

increases as the loading increases from 5 to 10 wt%Pt. As

the loading is further increased to 15 wt%Pt, the platinum
surface area remains essentially the same (.28 m versus
surface area remains essentially the same (.28 m versus








0.26 m2). This is not very surprising at the high loading

with inefficient impregnation. As the chemisorption results

for the 1 wt%Pt sample, which was prepared differently,

indicates, the impregnation procedures for the sample led to

a more efficient dispersion.

The equilibrium adsorption isobars obtained from exper-

iments were normalized with respect to the total BET surface

area and the results are shown in Figure 2 for the adsorbate

nitrous oxide and in Figure 3 for the adsorbate carbon diox-

ide. Given on the ordinate are Yi(T) for the pure platinum

and silica and vt / St for the supported catalyst samples.

The fractional catalyst surface area, S1 / St, can readily

be calculated from yi and vt / St given in the Figures 2 and

3 using Eqs. (6) and (10), provided b. are known.

As indicated earlier, the values of b. should be unity

if the adsorbate is indeed physisorbed. For the results

given in Figure 2, which is for the adsorbate of nitrous

oxide, the values of b. were assumed to be unity. The frac-
1
tional surface areas calculated from Eqs. (6) and (10) based

on the experimental results in Figure 2 are given in Table

2. The integrals in Eqs. (7) and (8) were calculated using

a four point Simpson's rule in the temperature range of -78

C to -45 0C. The points used in the numerical integration

are shown in Figures 2 and 3. The fractional catalyst sur-

face areas determined from hydrogen chemisorption are also

given in Table 2. The active surface area given in Table 1

was divided by the total surface area to obtain the percent

































s.-
0


4-

C
-

U
0a
-.3


Figure 2.


-70 -60 -50

T (oC)
Nitrous oxide selective physisorption on
platinum-silica catalyst (normalized
curves).









Table 2. Comparison between selective physisorption and hydrogen
chemisorption for platinum catalyst supported on silica.


%Pt Area %Pt Area
from from
Sample H2 chemisorption N20 physisorption


1 wt% Pt 5.6 8.6

5 wt% Pt 0.7 -0.9

10 wt% Pt 5.4 4.9

15 wt% Pt 5.7 6.9

*This is the actual calculated value which is within the
experimental error and should be considered as 0%.









o Silica
] Pt

.4 1% wt Pt
o 15% wt Pt

S10% wt Pt
5% wt Pt

0



ce
ar .3





CL
C-)











.1
-70 -60 -50
T (oC)

Figure 3. Carbon dioxide selective physisorption on platinum-silica
catalyst (normalized curves).


I








platinum area given in Table 2 under the heading H2 chem-

isorption.

The comparison made in Table 2 between chemisorption

and selective physisorption shows that for the higher load-

ing samples (5, 10, 15 wt%Pt), the selective physisorption

results for the percent platinum area compare well with the

chemisorption results, i.e. -0.9% (0%) versus 0.7%, 4.9%

versus 5.4%, and 6.9% versus 5.7% for 5, 10, 15 wt%Pt

catalysts respectively. The comparison for the 1 wt%Pt

catalyst (8.6% versus 5.6%) is not, however, as good as for

the higher loading catalyst samples. The crucial test for

the selective physisorption method was whether it can dis-

tinguish between .7% and 5.7% platinum surface area in a

reproducible manner, especially when the total surface areas

are as small as 0.037 and 0.324 m2/gcat, respectively. While

the selective physisorption results are not as accurate as

hydrogen chemisorption is for platinum-silica catalysts, the

comparison shows that it can still distinguish the differ-

ence between small variations of active catalyst surface

areas. This finding is significant in light of the fact that

there are at present no experimental methods that can give

such an accuracy for supported metal base oxide catalysts,

for which the selective physisorption is intended. This is

also true for some metal catalysts, as indicated by the

attempts of Miller and Lee (1984) with X-ray

diffraction/small angle scattering and SEM/TEM








techniques for the surface areas of potassium carbonate sup-

ported on carbon and silver supported on fused alumina.

If an independent method is available for comparison as

in the case of platinum catalyst, it is easy to determine

whether the chosen adsorbate is indeed inert to set b. equal

to unity since the comparison would allow this determina-

tion. For the supported catalysts for which an independent

method is not available, the definition of b. can be used to
1
determine whether the chosen adsorbate is suitable for the

selective physisorption method, i.e. whether the adsorbate

is inert enough such that b =l. According to the definition

of the b. (Eq. (10)), the adsorbate should not distinguish

between pure and dispersed solids when bi=1. Thus, vt/St

should be the same if b. = 1 whether it is for a physical

mixture or a supported catalyst as long as S1/St is the same

for both. For a chosen adsorbate for a given catalyst,

therefore, one can first assume that b. = 1 and then calcu-

late S1 / St from Eqs. (6), (11), and (12). Using this cal-

culated value of S1 / St, a physical mixture of the two com-

ponents constituting the supported catalyst can then be

prepared and vt / St obtained experimentally. A comparison

between vt/St thus determined and vt / St obtained experi-

mentally should reveal whether the chosen adsorbate is suit-

able for the supported catalyst. The better sensitivity

afforded by the use of Y-T relationship rather than e-T

relationship is illustrated in Figure 4 for the platinum

catalyst being considered. As was the case in Figure 1, The

























6

5D

40



34


2
10


-30 -60 -40 -20 0 20
T (oC)


Figure 4.


Comparison between the y-T cuves and
the e-T curves for the platinum silica
system.








use of the packing factor leads to a spread between the pure

components larger than that for the O-T relationship and

thus a more accurate determination of the fractional surface

area of the catalyst.

The results shown in Figure 3 for the adsorbate carbon

dioxide immediately reveal that carbon dioxide is not a

suitable adsorbate for the supported platinum catalyst. The

results also reveal that b. cannot be unity since the
1
curves do not lie between the Y1 and Y2 curves. Neverthe-

less, the results are quite intriguing in that carbon diox-

ide does not distinguish between pure platinum and pure sil-

ica and yet it distinguishes between supported catalysts of

different loadings as evident from the different Yt curves

shown in Figure 2. This finding suggests that the selective

physisorption of carbon dioxide could be used as a probe for

studying the platinum catalyst since the amount of chem-

isorbed hydrogen is the same regardless of dispersion

whereas the amount of selectively physisorbed carbon dioxide

per unit surface area of the catalyst depends on dispersion.

The work of Miller and Lee (1984) with carbon dioxide

and the results presented here with nitrous oxide fortify

the earlier contention that an adsorbate exhibiting a large

asymmetric directional polarizability is a good candidate

for the selective physisorption method. According to the

directional polarizabilities tabulated by Ross and Oliver




30


(1964), carbon disulfide, acetylene, and benzene should also

be considered for selective physisorption in addition to

nitrous oxide and carbon dioxide.













CHAPTER 4
APPLICATION OF THE SELECTIVE PHYSISORPTION METHOD
TO OXIDE CATALYSTS

As mentioned earlier, the main advantage of the method

of selective physisorption will be a more general applica-

tion to a wider variety of catalysts. If a suitable adsor-

bate is selected the new method should be equally applicable

to metal, metal oxide and metal compound catalysts. It

should be noted that although selective chemisorption has

been very successfully applied to many metal catalysts it

has not been effective on some other metal catalysts. The

situation is even worse in the case of metal compound and

metal oxide catalysts, where the chemisorption method has

had only limited success, and a considerable amount of

research is still being done to determine the fractional

catalyst surface area of these catalysts. In this chapter we

present the application of nitrous oxide selective phy-

sisorption to determine the fractional catalyst surface area

of two industrially important oxide catalysts namely: (1)

MoO3 / AI203, and (2) MoO3 / Si02 catalyst. Since no reli-

able independent method of measuring the fractional surface

area of these catalysts is available to date, a verification

of the results similar to the one carried out for the plati-

num silica catalyst using hydrogen chemisorption is not pos-

sible. Therefore the results of the selective physisorption








method for these catalysts will be verified by correlating

them with the cyclohexane dehydrogenation activity measured

with the help of a differential fixed bed reactor.

Experimental Apparatus

The apparatus used for the selective physisorption of

nitrous oxide on the oxide catalysts was identical to the

one used earlier for the metal catalyst except that the

thermal conductivity detector was replaced by a new Gow Mac

Temperature Regulated Cell Assembly. The new detector is

equipped with a temperature controller to control the tem-

perature of the detector and allows a much higher bridge

current to be used which renders the system much more sensi-

tive.

A differential bed plug flow reactor was used to deter-

mine the catalytic activity of the oxide catalyst for the

dehydrogenation of cyclohexane. A schematic diagram of the

reaction apparatus is given in Figure 5. The tubular reac-

tor was made of 12 mm quartz tube and had a length of 40 cm.

Spherical joints, also made of quartz, were placed at both

ends of the reactor tube to allow ease of loading of the

catalyst samples and cleaning of the reactor tube. A chromel

alumel thermocouple was placed on the outside wall of the

reactor tube and right above the differential catalyst bed,

to measure the reaction temperature. The reactor tube and

the thermocouple were wrapped tightly with a half inch wide

heating tape (Electrothermal Engineering Limited) which was

covered by a few layers of fiber glass insulating tape





















Bubble Flow
Meter


Figure 5. Schematic diagram of the differential reactor.







followed by about one and half inches of fiber glass insula-

tion blanket. The two ends of the heating tape were con-

nected to a variable power supply which was manually regu-

lated to control the reactor temperature at 425 0C. Each

catalyst sample was 0.5 to 1.0 gram in size and was secured

from both sides by about 1.5 cm long bed of pyrex wool.

The catalyst sample was loaded into the reactor tube to

ensure that all samples would be located at the same point

inside the reactor, right below the location of the thermo-

couple placed on the outside wall of the reactor tube. To do

this a small bed of pyrex wool was first pushed in through

the inlet side of the reactor with the help of a one foot

long glass rod. A weighed amount of catalyst sample was then

added to the reactor and secured from the other side with

the help of another piece of pyrex wool. This piece of

pyrex wool was pushed into the reactor tube to ensure a

snugly packed bed of catalyst. Excess force was not used in

order to avoid a very tightly packed catalyst bed which may

cause higher pressure drops and/or channeling of the feed

gas. After the catalyst had been loaded the spherical joints

at the two ends of the reactor tube were connected to the

respective gas lines and secured with the help of pinch

clamps. A very small amount of silicon grease (Dow Corning)

was applied to the joints to ensure leak free connections.

Nitrogen gas (Linde, Ultra High Purity, 99.999%) was

used as the inert carrier in the reactor. The nitrogen pres-

sure was set at 10 psig from the pressure regulator on the








gas cylinder and was never disturbed throughout the experi-

ments. A needle valve was used to regulate the inert car-

rier flow while the downstream by-pass valve was kept at

position 1 (reactant feed column by-pass) in the Figure 5.

The gas flow rate was measured with the help of a soap bub-

ble meter provided at the down stream end of the product gas

line. Once the carrier flow had been adjusted the needle

valve was not disturbed until all the catalysts had been

processed.

A specially designed reactant feed column, also shown

in Figure 5, was used to feed cyclohexane vapor to the reac-

tor. The bottom part of the 12 mm diameter pyrex column was

filled with liquid cyclohexane at room temperature. The

cyclohexane liquid level was brought up to a premarked level

before starting the reaction with each catalyst sample.

Nitrogen carrier gas was introduced in the feed column just

above the liquid surface to sweep the cyclohexane vapor

inside the feed column and exit at the top end of the reac-

tor. This arrangement was used to avoid excessive entrain-

ment of cyclohexane which might result if a bubble column

were used. Before this design was adopted a bubble column

was used to saturate the carrier gas with cyclohexane vapor.

The result was a very high entrainment of cyclohexane in the

feed gas. Every effort was made to avoid excessive entrain-

ment or to remove the entrained liquid from the feed gas but

with little success. The entrained liquid would settle in

the feed lines and the concentration of the reactant would








continue to increase and not stabilize even after several

hours of operation. The entrainment problem was eliminated

when the feed gas was introduced above the surface of the

liquid reactant as described above.

The feed column was immediately followed by a liquid

trap which was equipped with a teflon stopcock at the bot-

tom, to purge any liquid that may settle at the bottom of

the trap. The upper portion of the liquid trap was used as a

preheater which was heated by a coil of insulated nichrome

wire wrapped on the out side of the glass tube. In the

center of the preheater was a 2 mm thick glass fritt which

was used to ensure proper heating of the feed stream The

feed gas was then transmitted to the quartz reactor through

a 1/8 inch polyethylene tubing which was insulated with

fiber glass tape on the outside.

The product line from the differential reactor was

passed through a six port zero volume stainless steel gas

sampling valve (Supelco, model 2-2915) mounted on the sample

side of the gas chromatograph and then vented through a soap

bubble meter into the fume hood. A 0.25 ml stainless steel

sample loop (Supelco, model 2-2640) was used with the gas

sampling valve to take samples of the product stream for

analysis.

A Hewlett Packard model 5790A series gas chromatograph

equipped with a thermal conductivity detector was used for

the analysis of the reaction products. The gas connections

in the original chromatograph were altered to install a six








port zero volume stainless steel gas sampling valve

(Supelco, model 2-2915) on column A of the gas chromatograph

which was used for the sample gas. The gas sampling valve

was mounted right above the column oven and connected

directly to the chromatographic packed column bypassing the

injection port on the column A. This was done in order to

maintain the length of metal tubing between the gas sampling

valve and the packed column because a long tubing could

cause excessive peak broadening and thus introduce error.

The reaction products were separated with the help of a

1/8 in. X 6 foot stainless steel column (Supelco) packed

with di-n-decylpthalate on Chromosorb P (20 wt%) (Maggiore

et al., 1979). The column oven was programmed from 130 C to

160 C at a rate of 4 0C/min followed by a 6.5 min isother-

mal operation at 160 0C. The detector temperature and the

temperature of the injection port were kept at 250 0C. These

conditions were chosen after many trial runs to get sharp

separation of the product peaks at the fastest rate.

A Hewlett Packard model 3790A reporting integrator was

used to record the thermal conductivity cell response from

the chromatograph. The integrator was attached to the

chromatograph with the help of a remote starter cable in

order to synchronize the run of the chromatographic oven

with that of the integrator. Helium (Linde, Ultra High Pur-

ity, 99.999%) was used as the carrier gas in the chromato-

graph, and was further purified by passing through a molecu-

lar sieve trap (4 A, Davison) for the removal of water. The








helium flow rate was kept at 20 ml/min through both the

reference and the sample sides of the chromatograph. The

nitrogen gas used as the inert carrier in the differential

reactor was purified by passing through a bed of activated

carbon (American Scientific Products) followed by a bed of

13x molecular sieve (American Scientific Products) to remove

hydrocarbons and moisture respectively. The flow rate of the

gas in the reactor was about 10 ml/min.

Experimental Procedure

The chromatograph was started in accordance with the

instructions given in the operator's manual for the hewlett

packard 5790A series chromatograph. The helium supply pres-

sure was set at 50 psig from the gas cylinder and the flow

rate of the gas adjusted to 20 ml/min using the flow regula-

tor valves provided on the front panel of the chromatograph

and with the help of the soap bubble meter. After the car-

rier flow rate had stabilized the column oven was turned on

and its temperature brought to 130 OC mark. The detector

temperature was then set at 250 OC and the system allowed to

stabilize for eight to ten hours. In addition to this at

least one temperature programmed run was made on the chroma-

tograph before processing any samples of the product gas to

clean the packed columns. The detector current was turned on

at least two hours before conducting any experiments to

allow sufficient time for the system to stabilize. For over-

night shut downs the detector sensitivity was either lowered

or turned off to protect the detector elements from








excessive wear and tear. The thermal conductivity detector
_3
output was calibrated with the help of 1 x 10- ml injec-

tions of a standard mixture of cyclohexane and benzene.
-3
Another 1 x 10 ml injection of deionized distilled water

was used to calibrate the H20 peak.

Each catalyst sample was dried for six to ten hours

depending on whether it was powdered or in the form of pel-

lets. Without this drying time excessive amount of H20 would

be detected in the product stream. Visual observation deter-

mined that the moisture released from the catalyst, pores

upon heating in the reactor tube, would settle in the unin-

sulated spherical quartz joint at the end of the quartz

reactor and then continue to humidify the effluent gas

stream at a very slow rate. Nevertheless, six hours of dry-

ing was sufficient to remove all the moisture and the drying

was considered to be complete when no moisture was detected

in the product gas. It must be mentioned that the drying

was carried out at the reaction temperature and in an inert

helium environment. Also, since the drying temperature was

lower than the calcination temperature of 500 0C, it is

believed that this procedure did not effect other properties

of the catalyst. It is believed that the only effect of dry-

ing was the removal of moisture from the surface of the

catalyst.

After drying, the carrier bypass valve was switched to

position number 2 in the Figure 5. Thirty minutes were

allowed for the reaction to stabilize before taking the








first sample for analysis. It took 18.7 min for the chroma-

tographic oven to complete one run and return to the ready

state again. The following run was started as soon as the

oven returned to the ready state. A total of four runs was

made for each catalyst sample.

Catalyst Preparation

Five catalyst samples of moly-alumina catalyst with

0.1, 5, 10, 15, 25 wt%MoO3 per gram of catalyst were

prepared by impregnating high surface area alumina (Norton,

SA-6173; 1/16 in pellets) with an aqueous solution of

ammonium molybdate (Fisher; A-674). Deionized distilled

water (4.4 ml/5 galumina) was used to to prepare the impreg-

nating solution. The amount of ammonium molybdate used for

a particular catalyst was calculated to give the desired

loading of MoO3 on alumina. The actual amounts of the salt

and water used are given in the appendix. The aqueous solu-

tion prepared was just enough to completely saturate the

support material. Care was taken to ensure that very little

excess solution remained in the crucible after all the solu-

tion had been added, and at the same time no alumina pellets

were allowed to be left dry. The system was then allowed to

stand for 22 hours followed by 24 hours of drying in the

oven at 110 0C. The dried samples were transferred to a high

temperature furnace where the catalysts were calcined in air

for 17 hours at a temperature of 500 0C. After calcination

the samples were allowed to cool in air and then they were








transferred to glass sample bottles for storage. The total

weight of each sample prepared was 20 gm.

Five catalyst samples of moly-silica catalyst with 0.1,

5, 10, 15, 20 wt%MoO3 loading per gram of silica were also

prepared in the same way as the moly-alumina catalysts. High

surface area silica catalyst support (Alfa, 80396) was

impregnated with an aqueous solution of ammonium molybdate.

Deionized distilled water (12.5 ml/5 gsilica) was used to

prepare the impregnating solution. The actual amounts of

the salt and water used for each catalyst are given in the

appendix. Impregnation was allowed to continue for 24

hours. The catalyst seemed almost dry at the end of the

impregnation period but was dried further for 24 hours in an

oven at a temperature of 110 C. Calcination in air was car-

ried out for 18 hours in a furnace at 500 0C. The calcined

sample was crushed to obtain a uniform powder and was stored

in glass sample bottles.

Experimental Results And Discussion

Nitrogen BET was used to determine the surface area of

each catalyst using the continuous flow sorptometer. The

results indicate that molybdenum oxide is present in the

form of a monolayer on the surface area of alumina for up to

15wt%MoO3 loading catalyst. That is to say that the disper-

sion of MoO3 is very close to unity in the case of the 5,

10, and 15 wt%MoO3 loading catalysts. The 25 wt%MoO3 loading

moly-alumina catalyst is believed to carry MoO3 crystallites

and its behavior is therefore different from the lower








loading MoO3 loading moly-alumina catalysts. In the case of

the moly-silica catalysts the BET results suggest the forma-

tion of MoO3 crystallites for even the 5 wt%Mo03 loading

catalyst. A detailed discussion of the BET results for the

moly-alumina and moly-silica catalysts follows.

The BET surface areas of the moly-alumina catalysts are

given in Table 3. Figure 6 shows the decrease of total

catalyst surface area of different moly-alumina catalysts as

the MoO3 loading is increased. However, if the total surface

area of the individual catalysts is calculated per gram of

alumina, it is seen that it remains constant for the 5, 10,

and 15 wt%MoO3 catalysts, and then increases significantly

for the 25 wt%Mo03 catalyst. This observation that the total

surface area per gram of alumina for the 5, 10, and 15

wt%MoO3 loading moly-alumina catalysts remains constant, is

consistent with published data (Liu and Weller, 1980). It

conforms to the monolayer model for the moly-alumina

catalysts which suggests that in the unreduced catalyst

molybdenum oxide is present in the form of an epitaxial

monolayer on the surface of alumina. The monolayer model for

the unreduced moly-alumina catalyst is a widely accepted

model in the published literature (Massoth, 1973). Massoth,

for instance, has used butene chemisorption at 100 C on

freshly prepared 10 wt%MoO3 loading moly-alumina catalyst to

show a butene to molybdena ratio of 0.63 which compares well

with his theoretical value of 0.64 obtained by assuming that

all the molybdena were available on the surface for












Table 3. Nitrogen BET surface area of
moly-alumina catalyst.


Area Area

wt% Mo03 m2/gcat m2/galumina


0.1 164.0 166.0

5.0 172.1 181.0

10.0 163.0 181.0

15.0 152.7 180.0

25.0 154.0 204.0












205


195


0 5 10 15 20 25


Figure 6.


wt%MoO0
BET surface area of moly-alumina
catalyst.








adsorption. Massoth has also observed that only the 25

wt%MoO3 loading catalyst gave a sharp XRD peak for MoO3,

suggesting that crystalline moly is present only in the

higher loading moly-alumina catalysts. Stencel et al. (1983)

have studied MoO3/A1203 with the help of Raman Spectroscopy

and XRD. Their results also show that crystalline MoO3 is

present only in the higher loading moly-alumina catalysts.

Therefore we conclude that the increase in the BET surface

area (per gram of alumina) for the 25 wt%MoO3 loading moly-

alumina catalyst is due to the presence of crystalline moly

which contributes to the total surface area of the catalyst.

For the catalysts with up to 15 wt%MoO3 loading the molybde-

num oxide appears to form an epitaxial monolayer on the sur-

face of alumina.

The BET results for the moly-silica catalyst are given

in Table 4. It is seen that the total surface area of the

moly-silica catalysts decreases with increased MoO3 loading.

This can be explained by viewing the structure of the moly-

silica catalyst which is quite different from that of the

moly-alumina catalyst. No substantial evidence in the

literature was found which may suggest the formation of a

monolayer on the surface of silica. In fact, Garcia Fierro

et al. (1980) have shown the presence of MoO3 crystallites

on the surface of silica for a 13 wt%MoO3 loading moly-

silica catalyst with the help of a SEM micrograph. Even

though the total surface area of the moly-silica catalyst

remained constant for Garcia Fierro et al., in all the













Table 4. Nitrogen BET surface area of
moly-silica catalyst.


Area Area

wt% Mo3 m2/gcat m /gsilica


0.1 318.6 318.6

5.0 252.0 265.3

10.0 222.0 246.7

15.0 174.0 204.7

20.0 159.4 199.3


















300.






250






200 0






150






100
0 5 10 15
wt%MoO3
Figure 7. BET area of moly-silica catalyst.








catalysts the total surface area of the catalyst was much

less than the area that should have been contributed by the

amount of silica support present in the catalyst samples.

In addition to this the pore volume of their catalyst also

decreased from 1.34 cm3 /gm for silica to 1.03, 0.86, and

0.75 cm3 /gcat for the catalysts containing 4.8, 8.1 and

13.0 wt%MoO3, respectively. Thus it was concluded that the

Mo03 crystallites block the pores in silica and this results

in reduced surface area of the catalyst as the Mo03 loading

is increased. This is in contrast to the behavior exhibited

by the 25 wt% moly-alumina catalyst where the MoO3 crystal-

lites contributed to the total surface area of the catalyst.

However, it may be recalled that the alumina used in the

moly-alumina catalyst was in the form of pellets and its

pore size distribution may be different from the pore size

distribution of the silica used in the moly-silica catalyst.

Thus, while the moly crystallites may block the pores of

silica they may not block the pores of the alumina used in

the moly-alumina catalyst.

The results of nitrous oxide selective physisorption

indicate that nitrous oxide is not the ideal adsorbate for

the moly-alumina catalyst if the absolute value of the frac-

tional surface area is of interest. Also it is observed

that the values of the proportionality constants b. in Eq.

(10) cannot be unity for the nitrous oxide selective phy-

sisorption on the moly-alumina catalyst. Therefore, nitrous

oxide selective physisorption cannot be used to determine








the absolute value of the fractional catalyst surface area

of moly in the moly-alumina catalysts. Nevertheless, as

will be seen latter, ratios of the fractional surface areas

of different moly-alumina catalysts with up to 15 wt%MoO3

loading can still be evaluated with the help of nitrous

oxide selective physisorption data. In the case of the

moly-silica catalyst, it was found that nitrous oxide selec-

tive physisorption does not yield any useful information

regarding the fractional catalyst surface area. A detailed

discussion of these results follows.

The results of nitrous oxide selective physisorption at

-78 C on the MoO3/A1203 catalyst, and the Mo03/SiO2

catalyst are shown in Figures 8 and 9, respectively. Each

data point shown in these two figures represents the average

of the two experimental values obtained for each catalyst

sample being considered. The error analysis indicates that

the values of the packing factors may be +2% in error

(Appendix F). In light of this fact the differences between

the Yt values of the 5, 10, and 15 wt%MoO3 loading moly-

alumina catalysts (about 2% each) may not seem to be signi-

ficant. However, it was noted that the maximum scatter in

the experimental data for any individual catalyst never

exceeded 1.6% and in most cases was less than 1.0%. Also,

because the difference between the minimum and the maximum yt

values is much greater than 2%, we believe that the differ-

ences shown are indeed significant and the trend shown by

the data is real.
























0 9

C .095



.090:




.085




.080, r
0 5 10 15 20 25
wt%MoO3
Figure 8. Nitrous oxide selective physisorption
on moly-alumina catalysts.

















.100





.098





.096

0



c .094


o


S.092





.090





.088
.088---,-----,i- ,-----,--
0 5 10 15 20 25
wt%MoO
Figure 9. Nitrous oxide selective physisorption
on moly-silica catalyst.








In Figure 8 it is seen that the nitrous oxide packing

factor for the moly-alumina catalyst increases with

increased Mo03 loading with up to 15 wt%MoO3 loading, and

then decreases sharply for the 25 wt%MoO3 loading catalyst.

This behavior can be explained by viewing the different

structure of the higher loading catalyst. As it was dis-

cussed earlier, the molybdenum oxide in the 25 wt%MoO3

catalyst appears to be present in the form of Mo03 crystal-
2-
lites along with the one present as A12(Mo4 2-3 (Massoth,

1973). Since the packing factor for the pure Mo03 is much

less than the packing factor for pure alumina the MoO3 crys-

tallites in this catalyst may force the value of the packing

factor for the 25 wt%MoO3 loading catalyst to move towards'

the pure MoO3 value. Thus the packing factor for the 25

wt%MoOsub3 loading catalyst is much lower than those of the

lower loading moly-alumina catalysts. Because of this com-

plexity, the theory of selective physisorption cannot be

rigorously applied to the 25 wt%MoO3 catalyst.

For the lower loading moly-alumina catalysts, it is

clear that the packing factors for the dispersed catalysts

are higher than the packing factors for the pure components.

Therefore the b. in Eq. (10) cannot be unity for selective

physisorption of nitrous oxide on the moly-alumina catalyst.

In order to get any meaningful results from these data we

assume that the catalyst with 0.1 wt%MoO3 loading has negli-

gible fractional catalyst surface area. Considering that

the total surface area of the individual catalysts is of the







order of 170 m2 /gcat, this assumption does not seem to be

unrealistic. In that case then, the 0.1 wt%MoO3 catalyst

may be considered as the pure support with b2 equal to

unity. Therefore, from Eq. (10) we get


R1x1(T) = blY1(T)


(13)


R2x2(T) = Y2 (T).


(14)


Substituting these expressions in Eq. (5) for catalysts j

and k


S1
S








St k



Divide Eq. (15) by


S 1
S1

[ 1 ]
St k
Diid q.(5)b


vt [ 1
St j Y2(T)
blYl(T)
1 y2(T)





vt 1
b-tT.
1 [ bYl1 (T)
y2(T)


Eq. (16) to get



S t 1 1
St Y2 [ (T)
vt 1
S tk Y2(T)


and


and


(15)


(16)


(17)








which can be used to calculate the ratios of the fractional

catalyst surface areas of different catalysts. The results

of this calculation for the moly-alumina catalysts are given

in Table 5. Thus even in the case where b. are not unity it

is possible to at least evaluate the ratios of the frac-

tional surface areas of different wt% loading catalysts

using a catalyst with negligible wt% loading as pure sup-

port.

Theoretical values of fractional surface area ratios of

different moly-alumina catalysts, calculated by assuming a
2- 2-
complete monolayer formation of A12(MoO4 )3 and a MoO4

cross section of 25 A (Massoth, 1973), are also given in

Table 5. It is seen that the theoretical values do not com-

pare well with the experimental values. Also, if indeed the

monolayer model for the moly-alumina catalyst were valid,

and in light of the fact that the analysis of the BET

results does point towards the formation of an epitaxial

monolayer of molybdenum on the surface of alumina, the

results of the theoretical calculation should not be very

far from the actual values of the fractional surface area

ratios. On the other hand, it must be born in mind that the

theoretical calculations only represent an upper limit for

the fractional surface area ratios and the calculated values

are only hypothetical in nature. In any event, it is clear

that the values calculated from the nitrous oxide selective

physisorption are much lower than the theoretical values.







Table 5.


Ratios of fractional surface areas of
moly alumina catalysts.


Method (S1/St)15 (S1/St)10 (S1/St)15

(S1/St)5 (S1/St) (S1/St)O

N20 Physisorption 1.69 1.33 1.27

Monolayer Model 3.4 2.1 1.6








Since little is known about the stoichiometry of the

interactions between the selectively physisorbed gas and the

catalyst surface, we are unable to explain as to why the

nitrous oxide selective physisorption results are much lower

than the theoretically calculated values of the fractional

surface area ratios. Nevertheless, it is interesting to note

that the results of the two calculations follow the same

trend. It may also be pointed out that the fractional sur-

face area of a 15 wt%MoO3 moly-alumina catalyst obtained by

Parekh and Weller (1977), using low temperature oxygen chem-

isorption, was also about one fourth of the value that may

be obtained if a monolayer model were adopted. From this

discussion it is clear that more work needs to be done in

order to understand and explain the reason for the disagree-

ment between the results of the nitrous oxide selective phy-

sisorption and the theoretical results obtained by assuming

molybdenum monolayer on the surface of alumina. Since there

is no other independent method available to measure the

fractional surface area of the oxide catalysts, an absolute

verification of the nitrous oxide selective physisorption

method was not possible. Therefore, in order to make a

qualitative comparison, it was decided to compare the

cyclohexane dehydrogenation activity of the oxide catalysts

with the nitrous oxide packing factors obtained from the

selective physisorption of nitrous oxide.

The nitrous oxide selective physisorption results for

the moly-silica system do not yield any useful information








regarding the fractional surface area of the catalyst. At

best the packing factors for the moly-silica catalysts of

different wt%Mo03 loadings show a decreasing trend up to 15

wt%Mo03 loading catalyst, but due to the more complex nature

of this catalyst a better explanation cannot be offered at

present. The sudden increase in the value of the packing

factor for the 20 wt%Mo03 loading moly-silica catalyst is

not understood.

The cyclohexane dehydrogenation experiments were car-

ried out in a differential reactor and the results for the

moly-alumina catalyst system and the moly-silica catalyst

system are shown in the Figures 10 and 11, respectively.

The data shown in these two figures represents the run

number two for each catalyst being considered. It may be

recalled that four samples of the product gas were analyzed

for each catalyst sample. Since a differential reactor was

being used it is important to compare the activity of each

catalyst after it has been exposed to the same reaction con-

ditions for the same duration of time. The error analysis

indicates that the calculated values of the catalyst

activity may be +5% in error.

The rate of reaction was calculated from



r = F [ ] (18)
c A W


where F is the molar feed flow rate and AX and AW are the

differential conversion and the weight of the catalyst,








respectively. Cyclohexane, benzene and water were the only

components detected in the product stream. Since the reactor

was being operated in differential mode and the amount of

coking did not increase significantly with the increased

catalyst loading, it was neglected in the calculation of the

rate of reaction. The conversion was calculated by assuming

the reaction



C6H12 -> C6H6 + 3H2 (19)


only. The small amount of the H20 present in the product

stream may be attributed to the water attached to the

A12( Mo04 )3 species which could have been released upon

cyclohexane chemisorption on the surface of the catalyst. It

merits mention that the concentration of the H20 reduced

drastically in the first two runs and only trace amount was

detected in the subsequent runs, while the concentration of

cyclohexane and benzene remained constant in all the runs.

This observation fortifies our contention that the H20

present in the product stream does not represent a product

but is merely the water chemisorbed on the catalyst which is

released when cyclohexane chemisorption starts.

Figure 10 shows that no reaction was observed on pure

alumina or the 0.1 wt%Mo03 catalyst. After that the rate of

reaction increased steadily from 5 wt%Mo03 to 15 wt%Mo03

catalyst, and then the rate of increase of the reaction rate

slowed down towards the 25 wt%Mo03 catalyst. It is believed















1600


1400


1200


1000


800


600


400


200



0 5 10 15 20 25
wt%MoO3
Figure 10. Rate of cyclohexane dehydrogenation
on moly-alumina catalysts.
























700


600


500


400


300


200


100


0

Figure 11.


wt% Mo03
Rate of cyclohexane dehydrogenation
on moly-silica catalyst.








that this change in the rate of increase of the reaction

rate is primarily due to the change in the nature of the

catalyst surface as the loading increases from 15 to 25

wt%MoO3. It has been discussed earlier that MoO3 crystal-

lites may be present on the surface of the 25 wt%MoO3

catalyst. The present observation further strengthens our

belief that the 25 wt%MoO3 catalyst is different in nature

from the lower loading catalysts and, therefore, must be

treated separately.

The correlation between the nitrous oxide packing fac-

tors and the cyclohexane dehydrogenation activity of the

moly-alumina catalysts is shown in the Figure 12. It is

seen that the activity increases with the increase in the

value of the packing factor. This, however, is not the case

for the 25 wt%MoO3 catalyst but this catalyst has already

been shown to be an exception. Again, in spite of the rela-

tively small differences in the individual values for the

5, 10, and 15 wt%Mo03 catalysts (about 2% each), the corre-

lation is believed to be significant because of the good

reproducibility experienced in the selective physisorption

experiments and the overall trend of the data. Thus, the

correlation shown in the Figure 12 is believed to present a

valid partial verification of the nitrous oxide selective

physisorption results.

In the case of the moly-silica catalyst also, the

cyclohexane dehydrogenation activity is seen to increase

with increased MoO3 loading (Figure 11). Also, as in the



















1600
1600 25 wt%MoO3


1400 -


1200


1000


800
r-

* 600 .
(UI
4-)
S400


200


0 ,



.085 .090 .095 .100 .105
Y
Figure 12. Correlation between cyclohexane
dehydrogenation and nitrous oxide
packing factors for moly-alumina
catalyst.








case the moly-alumina catalyst, no reaction was observed on

pure silica or the 0.1 wt%MoO3 catalyst. The rate of reac-

tion increased steadily with increase in the Mo03 loading

and no shift in behavior was observed for the higher loading

catalyst. This observation along with the BET results shown

in the Figure 6 leads us to believe that all moly-silica

catalysts may be similar in nature in the sense that all

crystalline MoO3 may be present even in the lower loading

catalysts. No useful correlation between the activity of

cyclohexane dehydrogenation and the nitrous oxide selective

physisorption results could be obtained for the moly-silica

catalysts. Therefore we conclude that nitrous oxide selec-

tive physisorption does not provide any useful information

regarding the fractional catalyst surface area of the moly-

silica catalyst system.

In order to further study the moly-alumina catalyst

surface, the 10 wt%MoO3 catalyst was analyzed using X-ray

photoelectron spectroscopy. The survey scan for the 10

wt%MoO3 catalyst is shown in the Figure 13. The positioning

of the individual peaks on the binding energy scale can be

adjusted for the experimental work function by taking the Al

2p peak as the reference peak. According to the data given

in the handbook of X-ray photoelectron spectroscopy (Wagner

et al., 1979), the binding energy of the Al 2p peak for

A1203 is 74.1-74.3 eV, and the binding energy of the Al 2p
2-
peak for A12(MoO4 ) is 74.4 eV. Since only a fraction of
the total aluminum may be present as A 12(Mo04 2-)3, the Al 2p
the total aluminum may be present as A12(MoO4 )3, the Al 2p








peak is moved to 74.2 eV which is the average value of the

binding energy for A1203. This gives an experimental work

function of 4.85 eV. Using this experimental work function

value the binding energies of the Mo 3d, 0 Is, and C Is

peaks come out to be 232.85 eV, 531.4 eV, and 278.9 eV,

respectively. It is interesting to note that the binding

energy of the Mo 3d peak (232.85 eV) is very close to the
2-
the binding energy of the Mo 3d peak for A12(Mo042-)3 (232.8

eV). Also, since the binding energies for MoO3 and MoO2 are

less than 232.7 eV, we may conclude that Mo is present on
2-
the catalyst surface as A12(MoO4 2 3

A high resolution scan of the C Is peak is shown in the

Figure 14. The binding energy of the C Is peak suggests that

the carbon is present in the form of a carbide. A definitive

source of the surface carbon is not known. However, the car-

bon may partially be present due to carbon dioxide and car-

bon monoxide present in the atmosphere which may have chem-

isorbed on the catalyst surface upon exposure to the air.

It is quite likely that during calcination the chemisorbed

carbon oxides left some graphitic carbon on the catalyst

surface which in turn transformed into a carbide.

The surface composition of the 10 wt%Mo03 moly-alumina

catalyst obtained from the X-ray photoelectron spectroscopy

data is given in Table 6. It is seen that carbon represents

3.37 mass% of the total catalyst surface. However, it is not

clear whether the carbon is present on the catalyst surface

in the form of a thin monolayer or in the form of large








200


150





*100





50


0



0 200 400 600 800 1000 1200
Binding Energy (mV)
Figure 13. X-ray photoelectron spectroscopy survey scan for the
10 wt%MoO3 moly-alumina catalyst.










100.




80,




60
u

,


o



20




0... -1- i -- I

310 300 290 280 270
Binding Energy (eV)
Figure 14. High resolution scan of the carbon peak.









Tabl3 6. Composition of the 10 wt%MoO moly-alumina
catalyst as obtained from the x-ray photo-
electron spectroscopy data.


Quant Mass
Element Position Width Area Factor Ratio Cone %


Al 2p 74.2 3.45 591 0.29 7.35:1 55.72

Mo 3d 232.85 3.70 442 4.51 0.351:1 9.46

0 Is 531.4 3.50 1671 0.86 7.001:1 31.45

C Is 278.9 1.85 138 0.49 1:1 3.37








crystallites. Also it is not known as to how the presence of

this carbon may effect the nitrous oxide selective phy-

sisorption on the catalyst because similar carbon may be

present on all the catalyst samples as well as the two pure

components. Therefore, we recommend a thorough investigation

of these aspects when future work is undertaken. It will be

important to identify, with certainty, the source and form

of the surface carbon so that the effects of carbon on the

method of selective physisorption can be evaluated. Identif-

ication of the source may also aid in avoiding the carbon

deposit if that is deemed essential for the method of selec-

tive physisorption to be effective.

Summary

The method of selective physisorption for measuring the

fractional catalyst surface area of dispersed catalysts has

been applied to the oxide catalysts with moderate success.

Five moly-alumina catalysts of different MoO3 loading and

five moly-silica catalysts of different Mo03 loading were

prepared by impregnating high surface area alumina and high

surface area silica respectively, with aqueous solutions of

ammonium molybdate. The total surface area of each catalyst

sample was determined with the help of nitrogen BET method.

The results of the BET study indicate the formation of a

molybdenum oxide monolayer on the surface of alumina for the

lower MoO3 loading catalysts. That is to say that the

dispersion of molybdenum oxide on the surface of alumina is

unity for the 5, 10, and 15 wt%MoO3 catalysts, and is less








than unity for the 25 wt%MoO3 catalyst. This conclusion was

drawn from the observation that the total surface area of

the 5, 10, and 15 wtMoO3 loading moly-alumina catalysts

remains constant when it is calculated per gram of alumina

present in the catalyst. This behavior and the subsequent

conclusion is consistent with the view of most other

researchers in the literature who have studied the moly-

alumina system. The increase in the BET surface area of the

25 wt%MoO3 moly alumina catalyst is attributed to the pres-

ence of MoO3 crystallites on the surface of this catalyst.

The Mo03 crystallites have been confirmed to exist on the

higher Mo03 loading moly-alumina catalyst, with the help of

XRD and Raman spectroscopy.

The nitrous oxide selective physisorption was applied

to the moly-alumina catalyst. It was observed that the

nitrous oxide packing factors for the moly-alumina catalysts

increased with increased Mo03 loading for up to 15 wt%Mo03

loading catalyst, and then decreased sharply for the 25

wt%Mo03 loading moly-alumina catalyst. Also it was observed

that the nitrous oxide packing factors for the moly-alumina

catalysts with up to 15 wt%MoO3 loading are greater than the

nitrous oxide packing factors for the pure MoO3 and pure

alumina. Thus it was clear that nitrous oxide was not the

best suited adsorbate for the moly-alumina catalyst if the

absolute value of the fractional catalyst surface area is of

interest. Nevertheless, ratios of the fractional catalyst

surface areas of different catalysts with different wt%Ho03








loading can be computed if the 0.1 wt%Mo03 loading catalyst

were considered as the pure component. The underlying

assumption here was that the 0.1 wt%Mo03 loading moly-

alumina catalyst has negligible molybdenum oxide surface

area and therefore could be used as a reference. Thus the

fractional surface area ratios were computed to be 1.3, 1.6,

and 1.2 for 10 wt% to 5 wt%, 15 wt% to 5 wt%, and 15 wt% to

10 wt % Mo03 loading moly-alumina catalysts, respectively.

Since the surface characteristics of the 25 wt%Mo03 loading

moly-alumina catalyst are very different from the rest of

the catalysts it could not be included in the above calcula-

tion. The fractional surface area ratios mentioned above are

much lower compared to the theoretical values calculated

from the BET results assuming a monolayer of molybdenum

oxide on the surface of alumina. However, the results of the

two calculations follow the same trend.

The results of nitrous oxide selective physisorption

for the moly-alumina catalyst system were substantiated by a

correlation between the selective physisorption data and the

cyclohexane dehydrogenation activity for the moly-alumina

catalysts with up to 15 wt%Mo03 loading. The 25 wt%Mo03

catalyst was not included in the correlation because of its

different surface characteristics. The correlation is

believed to be significant in spite of the relatively small

differences in the individual packing factor values because

the scatter in the scatter in the experimental values was

even smaller. The overall trend of the data also strengthens








this belief. The moly-alumina catalyst was also analyzed

using X-ray photoelectron spectroscopy to further study the

surface of this catalyst. The results indicate that a sub-

stantial amount of carbon is present on the surface of the

catalyst in the form of a carbide. The exact source and

form of this surface carbon are not yet known with cer-

tainty. Also it is not known how this carbon may effect the

selective physisorption method. We recommend a more

thorough study of this phenomenon in order to develop the

method of selective physisorption further.

In contrast to the moly-alumina system the total sur-

face area of the moly-silica catalysts decreased with

increased MoO3 loading. This was so even when the BET sur-

face area of the catalysts was calculated per gram of silica

present in the catalysts. This observation and the results

of other researchers led us to believe that MoO3 crystal-

lites are present on the moly-silica catalyst even when the

catalyst loading is relatively low. The Mo03 crystallites

block the pores in the silica support and, therefore, the

total surface area of the catalysts decreases with increased

Mo03 loading. The nitrous oxide selective physisorption was

not successful in determining the fractional catalyst sur-

face area of the moly silica catalysts. At best a decreasing

value of the nitrous oxide packing factor with increased

MoO3 loading was observed for the moly-silica catalysts with

up to 15 wt%MoO3 loading. This trend reversed for the 20

wt%Mo03 loading catalyst and the reason for this behavior is




72



not understood. No correlation could be obtained between the

selective physisorption data and the cyclohexane dehydroge-

nation activity for the moly-silica catalyst system.














CHAPTER 5
CONCLUSIONS AND RECOMMENDATIONS

The method of selective physisorption for measuring the

fractional catalyst surface area of dispersed catalysts as

developed by Miller and Lee (1984) has been modified to

achieve increased sensitivity towards changes in the frac-

tional surface areas of catalysts with different loadings of

the dispersed material. It has been shown that the use of

packing factors (y) in place of fractional coverage (6)

affords a better sensitivity towards measuring the frac-

tional surface area of any catalyst. The theory has been

accordingly modified to incorporate packing factors instead

of fractional coverage. A few modifications have also been

suggested in the experimental method used for obtaining the

adsorption isobars of the selectively physisorbed gas. The

new method incorporates the use of more than one constant

temperature cold baths to obtain the adsorption isobars and

thus eliminates the need for the pseudo steady state assump-

tion that has been used by Miller and Lee (1984). Thus a

more dependable adsorption isobar is obtained in this

manner.

One important objective of this research was to demon-

strate the effectiveness of the method of selective phy-

sisorption in measuring the fractional surface area of








supported catalyst, and establish its credibility by using

it to determine the surface area of a standard catalyst sys-

tem. A standard catalyst system would be a one which would

provide an already established alternate method of determin-

ing its fractional surface area. Platinum dispersed on sil-

ica was chosen because the fractional surface area of

dispersed platinum can be determined very effectively with

the help of hydrogen chemisorption. Four samples of dif-

ferent wt%Pt loading on silica were prepared by impregnating

amorphous silica with an aqueous solution of chloroplatanic

acid. Both carbon dioxide and nitrous oxide were used to

independently determine the fractional surface area of the

dispersed platinum in the four catalysts. The results indi-

cate that carbon dioxide is not a suitable adsorbate for

measuring the fractional catalyst surface area of platinum

in the platinum silica catalyst, although it may provide

valuable insight into the said system. Nitrous oxide, on the

other hand, was found to be a good adsorbate to be used for

the selective physisorption on the platinum silica catalyst.

This observation also demonstrated that our earlier conten-

tion, that directional polarizability data can be used to

select an adsorbate for selective physisorption, was

correct.

The values of the fractional surface area of platinum

dispersed on silica determined from the selective physisorp-

tion of nitrous oxide were found to be in agreement with the

values obtained from the hydrogen chemisorption experiments.








It was shown that nitrous oxide selective physisorption is

sensitive to even small changes in the platinum surface area

as present in the supported catalyst.

In order to advance the method of selective physisorp-

tion to metal base oxide catalysts, nitrous oxide selective

physisorption was attempted on two industrially important

oxide catalysts, namely moly-alumina and moly-silica

catalysts. Five samples of each catalyst system were

prepared by impregnating high surface area alumina and high

surface area silica with aqueous solutions of ammonium

molybdate. The total surface area of each catalyst was

determined with the nitrogen BET method. The results indi-

cate that the total surface area of up to 15 wt%MoO3 loading

moly-alumina catalysts remained constant when calculated per

gram of the alumina present in the catalyst, and the total

surface area per gram of alumina of the 25 wt%MoO3 loading

moly-alumina catalyst increased significantly. This obser-

vation, along with the data and view of other researchers in

the published literature, leads us to the conclusion that

moly oxide forms a epitaxial monolayer on the surface of

alumina in the moly-alumina catalyst, with up to 15 wt%MoO3

loading, while crystalline MoO believed to be present in
is believed to be present in

the 25 wt%MoO3 loading moly-alumina catalyst along with the

moly oxide monolayer.

Nitrous oxide selective physisorption was only par-

tially successful when applied to the moly-alumina catalyst.

This is so, because the packing factors for the catalysts








with up to 15 wt%Mo03 loading were greater in value than the

packing factors for the pure Mo03 and pure alumina. This

clearly showed that b. for this system could not be unity.

Therefore absolute values of the fractional catalyst surface

areas of these catalysts could not be determined. Neverthe-

less, ratios of the fractional surface areas of different

moly-alumina catalysts can still be calculated by treating

the 0.1 wt%Mo03 loading catalyst as the pure support. The

underlying assumption here is that the fractional catalyst

surface area of the 0.1 wt%Mo03 catalyst is negligible com-

pared to the total surface area of the catalyst. The 25

wt%Mo03 catalyst was treated as an exception, and was not

included in the calculation being discussed here, because it

is believed to carry crystalline Mo03 on its surface which

makes the nature of its surface different from the nature of

the surface of the catalysts with lower catalyst loadings.

The results of this calculation do not compare well with the

results of theoretical calculations based on the monolayer

model for the moly-alumina catalysts.

An absolute verification of the nitrous oxide selective

physisorption results was not possible because no other

independent method was available to measure the fractional

surface area of the oxide catalyst. Instead a partial verif-

ication of results was made using a correlation between the

cyclohexane dehydrogenation activity and the nitrous oxide

packing factors for the moly-alumina catalysts. The correla-

tion is believed to be significant, in spite of the








relatively small differences between the individual packing

factor values, because the scatter in the experimental

values was even smaller. The overall trend of the data also

suggested that the correlation was indeed significant and

valid.

The BET results for the moly-silica catalyst system

indicate the presence of MoO3 crystallites on the surface of

silica which results in decreased total surface area of the

catalyst with increased Mo03 loading. Nitrous oxide selec-

tive physisorption on the moly-silica catalysts was not suc-

cessful, and it was determined that nitrous oxide is unsuit-

able for use as an adsorbate for the moly-silica catalyst.

No useful correlation could be obtained between the results

of nitrous oxide selective physisorption and the cyclohexane

dehydrogenation activity of the moly-silica catalysts.

The 10 wt%MoO3 moly-alumina catalyst was analyzed using

X-ray photoelectron spectroscopy to study the surface

further. It was observed that in addition to the oxygen,

molybdenum, and aluminum present on the surface of the

catalyst there was a substantial amount of carbon also

present on the surface. The exact source and form of the

surface carbon is not yet understood, but the carbon is

believed to come from carbon monoxide and carbon dioxide

present in the air. Also it is not known how this carbon

may effect the selective physisorption method. Therefore, it

is recommended that such surface analysis may be included in

future work in order to develop the theory of selective








physisorption method further. It will also be important to

correctly identify the source of the carbon present on the

surface of the catalyst so that it may be avoided if it

adversely effects the method of selective physisorption for

measuring fractional catalyst surface area.

It is the author's understanding that more work needs

to be done to better understand the nature of the interac-

tions between the selectively physisorbed gas and the solid

surface. Infra red spectroscopy may be able to offer some

insight into the state of the selectively physisorbed gases

which will help in establishing a better and more effective

criterion for the selection of a suitable adsorbate. Since

nitrous oxide selective physisorption was unable to deter-

mine the absolute values of the fractional surface areas of

the moly-alumina catalysts and nothing for the moly-silica

catalysts, we recommend more work in this area. A better

adsorbate needs to be selected to obtain more conclusive

results. The author suggests acetylene as a suitable candi-

date for use with the moly-alumina and moly-silica catalyst

systems.

An important application of the method selective phy-

sisorption may be in studying catalyst deactivation by cok-

ing. Preliminary results indicate a large difference between

the nitrous oxide packing factors for pure nickel and pure

graphite powders. The method may also be helpful in dif-

ferentiating between the catalyst deactivation caused by





79



coking and the catalyst deactivation caused by sulfur pois-

oning when the two occur simultaneously.














APPENDIX A
PREPARATION OF OXIDE CATALYSTS


Table 7. Preparation of moly-alumina catalyst.


Ammonium
wt% Mo03 Alumina Molybdate Water
(gm) (gm) (ml)


0.1 19.9786 0.0253 17.58

5.0 19.0033 1.2285 16.7

10.0 18.0000 2.4532 15.8

15.0 17.0044 3.6800 14.96

25.0 15.0005 6.1356 13.5












Table 8. Preparation of moly-silica catalyst.


Ammonium
wt% Mo03 Silica Molybdate Water
(gm) (gn) (ml)


0.1 3.9976 0.0045 50.0

5.0 3.8028 0.2467 47.5

10.0 3.6000 0.4912 45.0

15.0 3.3984 0.7376 42.5

20.1 3.2012 0.9858 40.0














APPENDIX B
NITROUS OXIDE SELECTIVE PHYSISORPTION DATA
FOR
PLATINUM-SILICA CATALYST


Table 9. Experimental conditions for nitrous oxide
selective physisorption on platinum-silica
catalyst.


Helium pressure (in H20) 25.1

Nitrous oxide pressure (in H20) 1.8

Nitrous oxide
Flow meter reading 30.0

Attenuation of desorption peak xl

Attenuation of calibration peak x4

Volume of calibration gas (cm3) 0.179

Area of calibration peak 99.3

Heater voltage (V) 15.

Total flow rate (cm3/min) 33.55

Helium flow rate (cm3/min) 30.185

Concentration chart speed (in/min) 2

Temperature chart speed (in/min) 0.5









Table 10. Experimental data
physisorption on


of nitrous oxide selective
platinum-silica catalyst.


K-Type
Peak Thermocouple Weight
Sample Run Area Reading (mV) (gm)


1.0 wt%



















5.0 wt%

















10.0 wt%


74.0

74.7

144.7

148.8

138.6

176.0

166.9

167.9

235.5

222.4

54.4

55.8

104.8

111.5

123.2

139.5

167.4

168.4

171.1

188.2

175.2


1.73

1.75

2.43

2.49

2.40

2.60

2.54

2.55

2.86

2.78

1.72

1.74

2.41

2.47

2.56

2.66

2.85

2.87

2.92

2.87

2.83


0.21395



















0.18115

















0.2012








Table 10--continued.


K-Type
Peak Thermocouple Weight
Sample Run Area Reading (mV) (gm)


15.0 wt%



















Si02


181.1

124.2

114.0

158.3

67.7

64.7

54.9

56.2

53.9

118.9

130.9

137.3

161.7

127.7

173.5

173.0

85.14

88.8

155.4

156.2

167.7

197.2


2.83

2.48

2.37

2.70

1.76

1.74

1.74

1.74

1.75

2.48

2.51

2.57

2.77

2.56

2.83

2.84

1.75

1.76

2.46

2.54

2.60

2.71


0.2185



















0.1971









Table 10--continued.


K-Type
Peak Thermocouple Weight
Sample Run Area Reading (mV) (gm)


222.3

225.9

226.9

232.1

46.1

43.4

72.8

68.4

75.7

74.7

76.0

71.8

72.4

73.6

84.5

85.4

85.9

86.4


2.85

2.85

2.88

2.91

1.73

1.70

2.47

2.41

2.67

2.63

2.60

2.56

2.51

2.57

2.82

2.85

2.86

2.90


0.39557













APPENDIX C
NITROUS OXIDE SELECTIVE PHYSISORPTION DATA
FOR
THE OXIDE CATALYSTS


Table 11. Experimental conditions.


Helium pressure 4.

Nitrous oxide pressure 3.5

Attenuation of desorption peak x16

Attenuation of calibration peak x4

Volume of calibration peak (cm3) 0.25

Area of calibration peak 38.5

Concentration chart speed (in/min) 4

Temperature chart speed (in/min) 4

Heater voltage (V) 8.7

Detector current (mA) 80

Detector temperature (oC) 60

Total flow rate (ml/min) 38.61

Helium flow rate (ml/min) 36.62





87





Table 12. Experimental data for nitrous oxide selective
physisorption on moly-alumina catalyst.


Sample Run Peak Temperature Weight
Area ( C) (gm)


5 wt%



10 wt%



15 wt%



25 wt%



0.1 wt%



Al 23


M203


Mo03


99.4

98.0

103.8

105.3

101.4

99.8

84.8

85.5

99.4

98.0

90.0

91.4


4.6

5.2


-78.

-78.

-78.

-78.

-78.

-78.

-78.

-78.

-78.

-78.

-78.

-78.


-78.

-78.


0.1462



0.1603



0.1609



0.1667



0.1626



0.1584




0.2632









Table 13. Experimental conditions for nitrous oxide selective
physisorption on moly-silica catalyst.


Helium pressure 3.4

Nitrous oxide pressure 3.0

Attenuation of desorption peak x32

Attenuation of calibration peak x4

Volume of calibration peak (cm3) 0.25

Area of calibration peak 52.4

Concentration chart speed (in/min) 4

Temperature chart speed (in/min) 4

Heater voltage (V) 8.7

Detector current (mA) 80

Detector temperature (oC) 60

Total flow rate (ml/min) 38.61

Helium flow rate (ml/min) 36.62





89





Table 14. Experimental data for nitrous oxide selective
physisorption on moly-silica catalyst.


Sample Run Peak Temperature Weight
Area ( C) (gm)


MoO3



Si02

0.1 wt%



20 wt%





15 wt%



10 wt%



5 wt%


8.7

8.7

97.8

91.7

91.6

47.7

48.2

47.9

49.7

49.5

66.3

66.0

67.8

68.5


-78.

-78.

-78.

-73.
-78.

-78.

-78.

-78.

-78.

-78.

-73.
-78.

-78.

-78.

-78.
-78.


0.2632



0.1263

0.1089



0.1207





0.1192



0.1245



0.1058














APPENDIX D
NITROGEN PHYSISORPTION RAW DATA







Table 15. BET data for pure platinum powder.


Weight = 0.33540 gm

Run

Helium pressure (in H20)

Nitrogen pressure (in H20)

Nitrogen flow meter

Attenuation
of desorption peak

Area of desorption peak

Attenuation
of calibration peak

Volume of
calibration gas (cm )

Area of calibration peak

Concentration
chart speed (in/min)

Temperature
chart speed (in/min)

P


1

25.0

0.

29.0


x2

45.5,55.9,48.3


x4


.5

168.8


1


.2

.075


2

25.0

0.

13.5


xl

82.3,79.7


x4


.2

65.8


3

22.1

5.0

13.5


x2

55.9,56.7


x4


.5

180.4


.2

.033


.2

.164


4

26.9

0.

0.


xl

63.1


x4


.5

174.0


1


.2

.004




Table 16. BET data for sintered silica.


Weight = 0.17400 gm

Run

Helium pressure (in H20)

Nitrogen pressure (in H20)

Nitrogen flow meter

Attenuation
of desorption peak

Area of desorption peak

Attenuation
of calibration peak

Volume of
calibration gas (cm3)

Area of calibration peak

Concentration
chart speed (in/min)

Temperature
chart speed (in/min)


1

22.0

4.2

48.0


x4

135.0


.5

186.0


2

23.7

2.0

27.0


x4

96.4,93.7


.5

176.0


.2

.170


.2

.073


3

26.

0.

4.0


x2

136.0,138


x4


.5

168.3


1


.2

.010







Table 17. BET data for 15.0 wt% platinum-silica catalyst.


Weight = 0.1925 gm

Run

Helium pressure (in H20)

Nitrogen pressure (in H20)

Nitrogen flow meter

Attenuation
of desorption peak

Area of desorption peak

Attenuation
of calibration peak

Volume of
calibration gas (cm3)

Area of calibration peak

Concentration
chart speed (in/min)

Temperature
chart speed (in/min)


1

22.5

5.0

51.5


x4

94.3,92.2


.5

166.6


.177


2

20.0

7.0

52.0


x4

117.9,118.1


.5

206.0


.2

.237


3

25.8

0.

6.0


x2

124.7,122.6


.5

168.2


.2

.021


4

24.7

1.0

30.5


x4

83.6,80.3


.5

169.5


.2

.079




Full Text

PAGE 1

FURTHER DEVELOPMENTS AND APPLICATION OF THE METHOD OF SELECTIVE PHYSISORPTION FOR MEASURING ACTIVE CATALYST SURFACE AREA By Irfan All Toor A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1985

PAGE 2

ACKNOWLEDGMENTS The author wishes to express his profound gratitude towards Dr. Hong H. Lee, chairman of the advisory committee, for suggesting this project and for his patience and continued guidance throughout this work. The author also wishes to acknowledge his gratitude towards Dr. Gar B. Hoflund, Dr. J. P. O'Connell, Dr. G.B. Westerman-Clark and Dr. E.R. Allen for serving on the advisory committee and for their valuable criticism ot the manuscript. Sincere appreciation Is also extended to Mr. Tracy Lambert, Mr. Ron Baxley and Mr. Rudi Strohschein for their skilled services In fabricating the experimental apparatus. Finally, the author wishes to thank his parents, brothers, sisters, wife and child for their patience and invaluable moral support without which the completion of this work would not have been possible. n

PAGE 3

TABLE OF CONTENTS ACKNOWLEDGMENTS KEY TO SYMBOLS ABSTRACT CHAPTERS I INTRODUCTION II MODIFICATIONS AND IMPROVEMENTS IN THE METHOD OF SELECTIVE PH YS I SORPTI ON Experimental Modi fi cat ions Theoretical Modi fi cat ions Ill APPLICATION OF THE SELECTIVE PHYSISORPTION METHOD TO METAL CATALYST Experimental Apparatus And Procedure... Catalyst Preparation Results And Discussion IV APPLICATION OF THE METHOD OF SELECTIVE PHYSISORPTION TO OXIDE CATALYST Experi mental Apparatus Experimental Procedure Catalyst Preparation Experimental Results And Discussion.... Summary V CONCLUSIONS AND RECOMMENDATIONS APPENDICES A PREPARATION OF OXIDE CATALYSTS Page i i V vi 8 8 15 16 18 19 31 32 38 40 41 68 73 80 m

PAGE 4

IV B NITROUS OXIDE SELECTIVE PHYS I SORPTION DATA FOR PLATINUM SILICA CATALYST 82 C NITROUS OXIDE SELECTIVE PHYS I SORPTI ON DATA FOR THE OXIDE CATALYSTS 86 D NITROGEN PHYS I SORPTI ON RAW DATA 90 E DIFFERENTIAL REACTOR DATA 110 F ERROR ANALYSIS OF THE PACKING FACTOR VALUES 113 LIST OF REFERENCES 114 BIOGRAPHICAL SKETCH 116

PAGE 5

b. 1 ^• KEY TO SYMBOLS proportionality constant in Eq. (10) integral defined by Eq. (7) or by Eq. (8) integral defined by Eq. (8) or by Eq. (12) volume of gas adsorbed on catalyst per unit surface area of the catalyst in the supported state volume of gas adsorbed on support per unit area of the support in the supported state catalyst surface area support surface area total surface area (S.+Sp) temperature lower limit for integration with respect to temperature upper limit for integration with respect to temperature volume of gas adsorbed on pure catalyst (i=l) or pure support (i=2) volume of gas adsorbed on the catalyst in the supported state volume of gas adsorbed on the support in the supported state Vj + Vj Subscript i = 1 for catalyst, 2 for support

PAGE 6

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy FURTHER DEVELOPMENTS AND APPLICATION OF THE METHOD OF SELECTIVE PHYSISORPTION FOR MEASURING ACTIVE CATALYST SURFACE AREA By Irfan Ali Toor May, 1985 Chairman: Dr. H.H. Lee Major Department: Chemical Engineering Modifications have been made in the method of selective phy si sorp ti on as developed by Miller and Lee (Miller, D.J., and H.H. Lee, AIChE J., 30, 84 (1984)). The new method makes use of packing factors of the selectively physisorbed gas instead of fractional coverage. The spread of packing factors versus temperature curves has been shown to be much greater than the spread of fractional coverage versus temperature curves. The method has been tested on a standard catalyst system of platinum supported on silica. The results have been compared with the results of hydrogen chemi sorpti on and it has been observed that they are in fair agreement. The application of the new method has been extended to metal oxide catalysts. The results indicate that nitrous oxide is not the best suited adsorbate for the moly-alumina catalyst, in the VI

PAGE 7

vn sense that the nitrous oxide selective phy si sorp ti on does not yield the absolute value of the fractional surface area of this catalyst. However, it can still be used to determine the ratios of the fractional surface areas of different molyalumina catalysts with up to 15 wtXMoO^ loading. The values of the fractional surface area ratios calculated from the nitrous oxide selective phy si sorp ti on results were much lower than the values computed from the BET results, assuming an epitaxial monolayer of moly oxide on the surface of alumina. It was shown that the cyclohexane dehydrogena ti on activity increases with the increase in nitrous oxide packing factors for the moly-alumina catalysts. The nitrous oxide selective physisorption was not successful on moly-silica catalyst.

PAGE 8

CHAPTER 1 INTRODUCTION Catalytic materials are prepared with the intention of maximizing the exposed surface area of the active material upon which the reaction may take place. Many important industrial catalysts are prepared by dispersing the active material on the surface of a porous material generally referred to as the support. Since the number of the cataly ti cal ly active sites is believed to be proportional to the total surface area of the dispersed material it is desirable to measure the surface area of the dispersed catalyst. Unfortunately, direct measurement of the surface area of the catalyst material dispersed in the pores of the support is not possible at the present time. Therefore various indirect techniques have been developed to characterize the dispersed catalyst. X-ray line broadening (Anderson , 1968), mossbaur spectroscopy (Bartholomew and Boudart, 1973), magnetization techniques (Hill and Selwood, 1949), and low angle scattering (Sinfelt, 1967), to mention just a few. Another important method of characterizing the porous catalysts is the physical and chemical adsorption of different gases to determine the surface area of the dispersed catalyst. Brunauer, Emmett, and Teller (1938) were the first to use the low temperature physical adsorption isotherms of 1

PAGE 9

various gases to measure the total surface area of the porous catalysts. This method, commonly known as the BET method, has found wide application in catalytic research all around the world. Brunauer and Emmett (1938, 1940) also studied the selective chemi sorption of various gases to determine the surface area of dispersed catalysts. Since that time, numerous attempts have been made to use the method of selective chemi sorpti on to determine the fractional surface area of oxide catalysts but they have met with limited success. The chemi sorpti on method differs from the phy si sorp ti on method in that, ideally, the chemical adsorption occurs only onto a particular component of the catalytic material. The method requires that the adsorbate form a monolayer of the chemisorbed atoms on the surface of the catalyst and that there exist a simple relationship between the number of molecules or atoms adsorbed and the number of surface atoms of the catalyst. Since most metals chemisorb small molecules like Hp, Op, and CO the chemi sorp ti on has been quite successful in characterizing many metal catalysts. For instance hydrogen chemi sorp ti on has been used very effectively to determine the fractional surface area of dispersed platinum catalysts (Spenedal and Boudart, 1960; Adler and Kearney, 1960). In the case of the metal oxide catalysts, however, the situation is further complicated because of two problematic points : (a) the lower interaction specificity of the possible gas/metal oxide system relative to that for

PAGE 10

gas/support system, and (b) the necessity to look for different gases for each supported metal oxide catalyst. Nevertheless, numerous attempts have been made to use selective chemi sorpti on of gases to characterize metal compound and metal oxide catalysts (Weller and Voltz, 1954 ; Segawa and Hall, 1983) but difficulty in the interpretation of the experimental results still remains. A series of articles has been published by Weller and associates (Parekh and Weller, 1977; Parekh and Weller, 1978; Srinivasan, Liu, and Weller, 1979; Liu, Yuan, and Weller, 1980; Garcia Fierro, Mendioroz, Pajares, and Weller, 1980; Liu and Weller, 1980) to discuss the low temperature oxygen chemi sorpti on for measuring the fractional surface area of molybdenum oxide dispersed on alumina but their results are at variance with current ideas concerning the formation of an epitaxial monolayer of molybdenum oxide on the surface of alumina. It has also been shown that in oxide catalysts the chemi sorpti on is site selective (Millman, Crespin, Crillo, Abdo and Hall, 1979; Vaylon and Hall, 1983) and therefore represents only a fraction of the actual surface area of the dispersed catalyst. From the above discussion it is clear that there was a need to look into new methods of gas adsorption of porous catalysts which may be applicable to a wider range of catalysts including metal and metal oxide catalysts. A new experimental method of measuring the active surface area of supported catalysts using selective physisorption of certain gasses has been put forward recently by

PAGE 11

Miller and Lee (1984). The basic idea here was that a monoor sub-monol ayer volume of a gas physisorbed on a two component solid as in a supported catalyst should be able to distinguish between the two different surfaces upon thermal desorption if the interactions between the gas and the surface are strong, say 3-6 Kcal/mole in terms of heats of adsorption. The desorption characteristics of gas coverage with temperature for the catalyst, the support, and the supported catalyst were used by Miller and Lee (1984) to calculate the fraction of the total surface area occupied by the catalyst. Because of the nature of phy si sorp ti on the selective phy si sorpti on should be applicable to any supported catalyst including oxides and metal compound catalysts provided a suitable adsorbate is used. This is in contrast to the chemi sorption method (Spenedal and Boudart, 1960; Adler and Kearney, 1960) which does not yield the catalyst surface area of some metal catalysts and the majority of metal compound catalysts because of the specific nature of chemisorption, although it yields valuable information of the sites active to a chemisorbing gas. In the original development (Miller and Lee, 1984), thermal desorption of selectively physisorbed carbon dioxide on the surface of potassium carbonate carbonate-carbon black mixtures was used to determine the fractional surface area of each component in the physical mixture. A qualitative comparison between the selective phy si sorpti on method and oxygen chemi sorp ti on was also given for supported silver

PAGE 12

catalyst. Nevertheless, no conclusive evidence was presented as to the effectiveness of the selective phy si sorpti on method as applied to the supported catalysts. Further the thermal desorption experiments and the data analysis were such that the method was effective only when catalyst loadings were high and the catalyst covered a significant portion of the total surface area of the support. In Chapter 2, we present refinements in the original method that would allow a more accurate determination of the catalyst surface area even when the catalyst covers a small portion of the support surface area. The new method involves the use of packing factors (y) in place of fractional gas coverage (e) to calculate the fractional surface area of the dispersed catalyst. The effectiveness of the selective phy si sorption method in measuring fractional catalyst surface area of dispersed catalyst is demonstrated in the Chapter 3. A catalyst system was sought that is amenable to a well established method of measuring the fractional surface area of the dispersed catalyst. Platinum dispersed on silica was chosen because hydrogen chemi sorption has been well established as a reliable method measuring the surface area of dispersed platinum and thus a valuable comparison of results could be made with the results of selective phy si sorpti on. Since an important advantage of the method of selective phy si sorpti on would be a more general and wider application to different types of catalysts, it was decided to test the

PAGE 13

new method on at least one industrially important metal oxide catalyst. The application of the selective physisorption method to moly-alumina catalyst is presented in Chapter 4. The results are correlated with cyclohexane dehydrogenation activity of the catalyst measured in a differential reactor. Conclusions and recommendations for future work are discussed in the Chapter 5.

PAGE 14

CHAPTER 2 MODIFICATIONS AND IMPROVEMENTS IN THE METHOD OF SELECTIVE PHYSISORPTION Selective phy si sorpti on utilizes the differences in the forces of interaction between different gas solid pairs. A detailed discussion of the nature of the phy si sorpti on forces and the kinetic models used to describe this phenomenon has been given by Miller (1982) and will not be presented here. Anyone interested in a more thorough understanding of the subject is referred to Brunauer (1943), Young and Crowell (1962), Clark (1970), Ricca (1972), Herz et al. (1982). The purpose of this research is to advance the theory of selective phy si sorption as applied to the measurement of the fractional surface area of supported catalyst. A few modifications have been made in the experimental method and the method of data analysis which render the methods of selective phy si sorp ti on more sensitive to changes in the fractional surface area of the dispersed catalyst. These modifications and improvements will be discussed in the following paragraphs. The main contribution of this study, however, is the demonstration of the effectiveness of the selective phy si sorpti on in determining the fractional surface area of supported metal catalysts (Chapter 3), and its extension to oxide catalysts (Chapter 4).

PAGE 15

Experimental Modifications In the original development. Miller and Lee (1984) utilized the differences in the gas coverage ( e) versus temperature (T) relationships of the monoor sub-monol ayer volume of a physisorbed gas to determine the fraction of the total surface area covered by the catalyst. In order to generate the thermal desorption isobars it was assumed that a pseudo steady state exists during the transient thermal desorption of the monoor sub-monol ayer of the physisorbed gas. Therefore a single thermal desorption was sufficient to obtain a e-T relationship (Miller and Lee, 1984). The pseudo steady state assumption was avoided in this study and instead more than one constant temperature cold baths were used to determine the amount of gas adsorbed at each temperature in order to generate a more accurate e-T relationship. The motivation is that the total desorption will provide a more accurate determination of the amount of gas adsorbed and thus a more accurate adsorption isobar will be achieved which in turn will enhance the accuracy with which the catalyst surface area is determined. Theoretical Modifications While Miller and Lee (1984) have used the e-T relationship to determine the fractional surface area of the supported catalyst, it was found that the use of packing factor (which is the amount of gas adsorbed per unit surface area of the supported catalyst) in place of 6 would allow a more accurate determination of the fractional surface area of the

PAGE 16

supported catalyst. For instance, consider Figure 1, which compares the 9-T curves of the pure components of potassium carbonate and carbon black mixture, obtained by Miller and Lee (1984), with the spread of the Y (the packing factor expressed as volume of gas per unit N^ BET area) versus T curves generated from the same data. It is clear that the spread of the y-T curves is much greater than the spread of the e-T curves and therefore the former relationship will provide a better sensitivity towards the determination of the fractional surface area of the dispersed catalyst. In light of the above development it was necessary to develop an expression for the determination of the fractional surface based on the packing factors rather than 9 in order to take advantage of the greater sensitivity afforded by the packing factor. If we let V, and vbe the volumes of the submonolayer of a gas adsorbed on the surfaces 1 (catalyst) and 2 (support) in the supported catalyst then the total volume of the gas adsorbed on the supported catalyst v. is simply the sum of v^ and v^: v^(T) = v^(T) + V2(T). (1) Further if we let x.(T) be the fraction of the surface area of the surface i that is covered by the adsorbate at the temperature T, the total surface area of the supported catalyst, S^, is given by

PAGE 17

10 K2C03 .1.4 J .2 .1 .n •80 -60 20 -40 -20 T (°C) figure 1. Comparison between y-T curves and 6-T curves for the potassium carbonate-carbon black system. CD s> o o 8 _ (T3 c o •^ 4-> o

PAGE 18

11 s = ^^ + ^ J(2) where the factor R. is given by cm adsorbate on surface i 2 m surface area of surface i (3) Since v. cannot be measured directly, but v. can be measured using thermal desorption, we rearrange Eq. (2) with the help of Eq. (1) to get v^(T) h

PAGE 19

12 integrated over the entire temperature range of interest and the result will still be valid. This is done in order to take advantage of the entire adsorption isobar. Also the integration cancels out the scatter in the experimental data and the calculated value will be the average value of the fractional surface area of the dispersed catalyst, representing the temperature range of interest of the adsorption isobar. Multiplying S, / S by the numerator of Eq. (5), integrating, and rearranging yield the following final result (6) where [ 1 ^ ^t '^2^2 ] dT (7) and [ 1 R P X ~ ] dT, (8) Here T^. and T^ are the temperatures chosen for integration in the temperature range of interest. The values of R^.x^. can readily be obtained experimentally for the two pure components constituting the supported

PAGE 20

13 catalyst. The definitions of R. and x., when applied to the pure components, yield [ R.x^. (T) ] = ( N BET area ) . ^ 2 1 , pure Z ^^(T) (9) where U. is the sub-monol ayer volume of the gas adsorbed on pure catalyst (i=l) and pure support (i=2) surfaces at temperature T. It should be noted that U. is different from v.. 1 1 For the platinum catalyst supported on silica (Chapter 3), for instance, the volume of gas adsorbed on the platinum particles is U. whereas the volume of gas adsorbed on the platinum particles dispersed in the supported catalyst is V,. If we assume that the quantity R.x. for the surface unsupported state is proportional to that for the surfaces in the pure state, we have R^.x.(T) = b. Y^-(T) (10) where b. is the proportionality constant. However, if the adsorbate is indeed physisorbed it will not distinguish between pure and dispersed states of the catalyst and therefore the constants b. should assume a value of unity. For such an adsorbate, R.x. is equal to y. and all the quantities in Eqs. (7) and (8) can readily be obtained from experiments, thus allowing the calculation of the fractional

PAGE 21

14 catalyst surface area from Eq. (6) because Eqs. (7) and (8) reduce to (1 ) dT (11) 1 J (1 ) dT (12) where Y^ = v / S .

PAGE 22

CHAPTER 3 APPLICATION OF THE SELECTIVE RHYS I SORPTI ON METHOD TO METAL CATALYSTS One shortcoming of the work of Miller and Lee (1984) was that no conclusive evidence was provided as to the effectiveness of the selective phy si sorpti on method as applied to the dispersed catalysts. They used the selective physi sorpti on of carbon dioxide at -78 C to determine the fractional catalyst surface area of silver catalyst dispersed on alumina and compared the results with those of oxygen chemi sorpti on . However, the stoi chiometry of oxygen chemi sorpti on on dispersed silver catalyst is not fully understood (Miller, 1982) and is not a well established method of measuring the fractional surface area of silver catalyst. Furthermore the comparison made by the original authors was only qualitative in nature, and therefore does not furnish a positive verification of the surface area calculated using thermal desorption of the selectively physisorbed carbon dioxide. Scanning electron microscopy was also attempted to determine the surface area of dispersed silver but the results were inconclusive. The other catalyst system studied by Miller and Lee (1984) was potassium carbonate dispersed on carbon black. Again no independent measurement of potassium carbonate surface area could be made and thus no comparison was made. 15

PAGE 23

16 Physical mixtures of carbon black and potassium carbonate were then used to partially verify the results but such a comparison can only be of a limited value. In order to provide a positive evidence that the selective phy si sorpti on method can be used effectively to determine the fractional surface area of dispersed metal catalyst, a catalyst system was sought that would provide an already established alternate method of independently measuring the fractional surface area of the catalyst so that the results of the selective phy si sorpti on method could be verified. Platinum supported on silica was chosen because it meets these conditions perfectly. Hydrogen chemi sorpti on has been studied extensively (Spenedal and Boudart, 1960; Adler and Kearney, 1960; Wanke, Lotochinski and Sidwell, 1981; Sarakany and Gonzalez, 1982) and is a well established method of measuring the fractional surface area of platinum catalyst. Therefore it was decided to determine the fractional surface area of platinum in pi ati numsi 1 i ca catalyst using the new method and to verify the results by making a comparison with the platinum surface area determined from hydrogen chemi sorpti on . In this manner, it would be possible to demonstrate the effectiveness of the selective physisorption method in determining the fractional surface area of metal catalysts. Experimental Apparatus And Procedure The gas adsorption experiments in this study have been carried out with the help of a Perkin Elmer continuous flow

PAGE 24

17 sorptometer modified for thermal desorption experiments. A detailed description of this apparatus has been given by Miller (1982) and will not be presented here. The constant temperature cold baths that were used for low temperature adsorption were made from liquid nitrogen, acetone-dry ice, chloroform-dry ice, and chl orobenzene-dry ice mixtures to obtain -196 °C, -78 °C, -61 °C, and -45 °C (Phipps and Hume, 1968) respectively. The sample cell containing the catalyst sample (Miller, 1982) was immersed in the cold bath and the system was allowed to equilibrate. Equilibrium was determined by a constant reading of the temperature recorder for a period of 4-5 minutes, and a stable response from the thermal conductivity cell detector. After the equilibrium had been reached the cold bath was removed instantaneously and the sample was heated with the help of an external heating coil, to completely desorb the gas. The response of the thermal conductivity cell was recorded using an integrator recorder. A minimum of two readings were taken at each adsorption temperature. A 0.5 ml pulse of the adsorbate gas was used to calibrate the response of the thermal conductivity cell. All samples were processed one after the other within a period of 24 hrs. to ensure minimal fluctuations in the gas flow rate. The gas flow rates were measured with a soap bubble meter after the last sample had been processed, as detailed in the reference.

PAGE 25

18 The hydrogen chemi sorp ti on experiments were carried out using the dynamic pulse method (Wanke, Lotochinski and Sidwell, 1981). A detailed description of the method and the experimental apparatus is available in the reference and need not be presented here. Pulses of 0.1 ml of hydrogen (Airco Grade 5.5, 99.9995%) were used for the chemi sorpti on at 100 C. Each gas was purified by first passing through a bed of 4 °A molecular sieve (Davison Chemical, Grade 5.3, 4-8 mesh beads) and then through a bed of copper. In the BET and the low temperature adsorption experiments, helium (Airco Grade 4.5, 99.995%) was used as the carrier gas with nitrogen (Linde, Ultra High Purity, 99.999%), carbon dioxide (Airco Grade 4, 99.99%) or nitrous oxide (Matheson, Ultra High Purity, 99.999%) as the adsorbate. Each gas was purified by first passing through a 4 °A molecular sieve (Davison Chemical Grade 5.3, 4-8 mesh beads) which was followed by a cold trap. The cold trap consisted of liquid nitrogen in a dewer flask when nitrogen was used as adsorbate, or acetone-dry ice slush in a dewer flask when nitrous oxide of carbon dioxide was used as the adsorbate. Catalyst Preparation Platinum-silica catalysts were prepared by impregnating powdered silica (Alpha, 99.5% SiO^, -400 mesh amorphous) with chloroplatanic acid solution. A stock solution of chl oroplatani c acid was prepared by dissolving 5.1 gm of chloroplatanic acid (Alpha) in 100 ml of distilled water. Impregnation was carried out by adding 11.5, 24.5 ml, and

PAGE 26

19 36.0 ml of the stock solution to three 5 gm samples of sintered silica to prepare 5 wt%, 10 wt%, and 15 wt%Pt catalysts respectively. Each sample was allowed to stand for a 24-hr period for impregnation. Intermittent stirring was continued during this 24-hr period. The samples were then dried in an oven at 120 °C for 24 hrs, and then reduced in flowing hydrogen at 500 °C for 17 hrs. Another catalyst sample with 1 wt%Pt loading was prepared in the following manner. A calculated amount of chloropl atani c acid which gives 1 wt%Pt loading on 10 grams of silica was dissolved in 5 grams of distilled water. The resulting solution was added to 10 grams of amorphous silica (Alpha, gg.SXSiOp, -400 mesh amorphous)to obtain a thick slurry. This was allowed to stand for a 24-hr period for impregnation. During this time, intermittent stirring was continued to maintain a uniform slurry. At the end of the 24-hr period, the impregnated slurry was dried at 120 °C for another 24 hrs. This process was followed by 16 hrs of reduction at 500 C in a stream of flowing hydrogen. No pretreatment was carried out to increase the dispersion of any catalyst. Results And Discussion The BET and hydrogen chemi sorp ti on results are summarized in Table 1. The BET total surface area for each sample was determined using four data points at four different pressures. A one-to-one hydrogen-to-platinum correspondence was used to convert the amount of hydrogen chemisorbed on

PAGE 27

20 K

PAGE 28

21 the active platinum surface area of the supported catalyst, 19 2 using a value of 1.12 x 10 Pt sites/m (Spenedal and Boudart, 1960). It is interesting to note in Table 1 that the total BET surface per gram of supported catalyst decreases with increasing platinum loading. This behavior could be due to a growing number of platinum crystallites as the platinum loading is increased. The large crystallites may block micropores and thereby cause a reduction in the total surface area of the supported catalyst. The chemi sorpti on results reported in Table 1 (equi valently in Table 2) need some comments. As indicated earlier, the impregnation technique for the 1 wt%Pt catalyst was different from those for 5, 10, 15 wt%Pt samples. As apparent from the table, inefficient impregnation for the higher loading catalysts resulted in low platinum dispersion. Because of the unusual behavior, the chemi sorp ti on experiments were repeated only to confirm the reported values in Table 2. This apparent drawback, however, was instrumental in demonstrating that selective phy si sorpti on can distinguish small differences in fairly small surface areas , which is an important feature because the technique was shown earlier (Miller and Lee, 1984) to be effective only for large surface areas. One can see for the higher loading sample catalysts that the catalyst surface area increases as the loading increases from 5 to 10 wt%Pt. As the loading is further increased to 15 wt%Pt, the platinum 2 surface area remains essentially the same (.28 m versus

PAGE 29

22 0.26 m ) . This is not very surprising at the high loading with inefficient impregnation. As the chemi sorpti on results for the 1 wt%Pt sample, which was prepared differently, indicates, the impregnation procedures for the sample led to a more efficient dispersion. The equilibrium adsorption isobars obtained from experiments were normalized with respect to the total BET surface area and the results are shown in Figure 2 for the adsorbate nitrous oxide and in Figure 3 for the adsorbate carbon dioxide. Given on the ordinate are Y--(T) for the pure platinum and silica and v. / S^ for the supported catalyst samples. The fractional cata lyst surface area, S, / S^, can readily be calculated from y. and v. / S^ given in the Figures 2 and 3 using Eqs. (6) and (10), provided b. are known. As indicated earlier, the values of b. should be unity if the adsorbate is indeed physisorbed. For the results given in Figure 2, which is for the adsorbate of nitrous oxide, the values of b. were assumed to be unity. The fractional surface areas calculated from Eqs. (6) and (10) based on the experimental results in Figure 2 are given in Table 2. The integrals in Eqs. (7) and (8) were calculated using a four point Simpson's rule in the temperature range of -78 °C to -45 °C. The points used in the numerical integration are shown in Figures 2 and 3. The fractional catalyst surface areas determined from hydrogen chemi sorpti on are also given in Table 2. The active surface area given in Table 1 was divided by the total surface area to obtain the percent

PAGE 30

23

PAGE 31

24 c

PAGE 32

25 o O 4->

PAGE 33

26 platinum area given in Table 2 under the heading H^ chemi sorption. The comparison made in Table 2 between chemi sorp ti on and selective phy si sorpti on shows that for the higher loading samples (5, 10, 15 wt%Pt), the selective phy si sorp ti on results for the percent platinum area compare well with the chemisorpti on results, i.e. -0.9% (0%) versus 0.7%, 4.9% versus 5.4%, and 6.9% versus 5.7% for 5, 10, 15 wt%Pt catalysts respectively. The comparison for the 1 wt%Pt catalyst (8.6% versus 5.6%) is not, however, as good as for the higher loading catalyst samples. The crucial test for the selective physi sorp ti on method was whether it can distinguish between .7% and 5.7% platinum surface area in a reproducible manner, especially when the total surface areas 2 are as small as 0.037 and 0.324 m /gcat, respectively. While the selective phy sisorpti on results are not as accurate as hydrogen chemi sorp ti on is for platinum-silica catalysts, the comparison shows that it can still distinguish the difference between small variations of active catalyst surface areas. This finding is significant in light of the fact that there are at present no experimental methods that can give such an accuracy for supported metal base oxide catalysts, for which the selective phy si sorp ti on is intended. This is also true for some metal catalysts, as indicated by the attempts of Miller and Lee (1984) with X-ray diffraction/small angle scattering and SEM/TEM

PAGE 34

27 techniques for the surface areas of potassium carbonate supported on carbon and silver supported on fused alumina. If an independent method is available for comparison as in the case of platinum catalyst, it is easy to determine whether the chosen adsorbate is indeed inert to set b. equal to unity since the comparison would allow this determination. For the supported catalysts for which an independent method is not available, the definition of b. can be used to determine whether the chosen adsorbate is suitable for the selective phy sisorption method, i.e. whether the adsorbate is inert enough such that b.=l. According to the definition of the b. (Eq. (10)), the adsorbate should not distinguish between pure and dispersed solids when b.=l. Thus, v /S. should be the same if b. = 1 whether it is for a physical mixture or a supported catalyst as long as S,/S. is the same for both. For a chosen adsorbate for a given catalyst, therefore, one can first assume that b. = 1 and then calculate S^ / S^ from Eqs. (6), (11), and (12). Using this calculated value of S, / S., a physical mixture of the two components constituting the supported catalyst can then be prepared and v. / S. obtained experimentally. A comparison between v./S. thus determined and v. / S. obtained experimentally should reveal whether the chosen adsorbate is suitable for the supported catalyst. The better sensitivity afforded by the use of Y-T relationship rather than e-T relationship is Illustrated in Figure 4 for the platinum catalyst being considered. As was the case in Figure 1, The

PAGE 35

28 > 4 ° o rtJ 2 ? -30 -60 -40 -20 20 T (°C) Figure 4. Comparison between the y-T cuves and the e-T curves for the platinum silica system.

PAGE 36

29 use of the packing factor leads to a spread between the pure components larger than that for the 6-T relationship and thus a more accurate determination of the fractional surface area of the catalyst. The results shown in Figure 3 for the adsorbate carbon dioxide immediately reveal that carbon dioxide is not a suitable adsorbate for the supported platinum catalyst. The results also reveal that b. cannot be unity since the curves do not lie between the Y-, and y^ curves. Nevertheless, the results are quite intriguing in that carbon dioxide does not distinguish between pure platinum and pure silica and yet it distinguishes between supported catalysts of different loadings as evident from the different y curves shown in Figure 2. This finding suggests that the selective phy sisorpti on of carbon dioxide could be used as a probe for studying the platinum catalyst since the amount of chemisorbed hydrogen is the same regardless of dispersion whereas the amount of selectively physisorbed carbon dioxide per unit surface area of the catalyst depends on dispersion. The work of Miller and Lee (1984) with carbon dioxide and the results presented here with nitrous oxide fortify the earlier contention that an adsorbate exhibiting a large asymmetric directional pol ari zabi 1 i ty is a good candidate for the selective phy si sorp ti on method. According to the directional pol arizabi 1 i ties tabulated by Ross and Oliver

PAGE 37

30 (1964), carbon disulfide, acetylene, and benzene should also be considered for selective phy si sorption in addition to nitrous oxide and carbon dioxide.

PAGE 38

CHAPTER 4 APPLICATION OF THE SELECTIVE PHYS I SORPTI ON METHOD TO OXIDE CATALYSTS As mentioned earlier, the main advantage of the method of selective phy si sorpti on will be a more general application to a wider variety of catalysts. If a suitable adsorbate is selected the new method should be equally applicable to metal, metal oxide and metal compound catalysts. It should be noted that although selective chemi sorpti on has been very successfully applied to many metal catalysts it has not been effective on some other metal catalysts. The situation is even worse in the case of metal compound and metal oxide catalysts, where the chemisorpti on method has had only limited success, and a considerable amount of research is still being done to determine the fractional catalyst surface area of these catalysts. In this chapter we present the application of nitrous oxide selective physisorption to determine the fractional catalyst surface area of two industrially important oxide catalysts namely: (1) MoO^ / AKO-, and (2) MoO^ / SiOp catalyst. Since no reliable independent method of measuring the fractional surface area of these catalysts is available to date, a verification of the results similar to the one carried out for the platinum silica catalyst using hydrogen chemi sorpti on is not possible. Therefore the results of the selective phy si sorpti on 31

PAGE 39

32 method for these catalysts will be verified by correlating them with the cyclohexane dehydrogenati on activity measured with the help of a differential fixed bed reactor. Experimental Apparatus The apparatus used for the selective phy si sorp ti on of nitrous oxide on the oxide catalysts was identical to the one used earlier for the metal catalyst except that the thermal conductivity detector was replaced by a new Gow Mac Temperature Regulated Cell Assembly. The new detector is equipped with a temperature controller to control the temperature of the detector and allows a much higher bridge current to be used which renders the system much more sensitive. A differential bed plug flow reactor was used to determine the catalytic activity of the oxide catalyst for the dehydrogenati on of cyclohexane. A schematic diagram of the reaction apparatus is given in Figure 5. The tubular reactor was made of 12 mm quartz tube and had a length of 40 cm. Spherical joints, also made of quartz, were placed at both ends of the reactor tube to allow ease of loading of the catalyst samples and cleaning of the reactor tube. A chromel alumel thermocouple was placed on the outside wall of the reactor tube and right above the differential catalyst bed, to measure the reaction temperature. The reactor tube and the thermocouple were wrapped tightly with a half inch wide heating tape (Electrothermal Engineering Limited) which was covered by a few layers of fiber glass insulating tape

PAGE 40

33 s. o U ns 0) •^ +j O) s«4T3 0) o (J +J «a E (U .c u CO in

PAGE 41

34 followed by about one and half inches of fiber glass insulation blanket. The two ends of the heating tape were connected to a variable power supply which was manually regulated to control the reactor temperature at 425 °C. Each catalyst sample was 0.5 to 1.0 gram in size and was secured from both sides by about 1.5 cm long bed of pyrex wool. The catalyst sample was loaded into the reactor tube to ensure that all samples would be located at the same point inside the reactor, right below the location of the thermocouple placed on the outside wall of the reactor tube. To do this a small bed of pyrex wool was first pushed in through the inlet side of the reactor with the help of a one foot long glass rod. A weighed amount of catalyst sample was then added to the reactor and secured from the other side with the help of another piece of pyrex wool. This piece of pyrex wool was pushed into the reactor tube to ensure a snugly packed bed of catalyst. Excess force was not used in order to avoid a very tightly packed catalyst bed which may cause higher pressure drops and/or channeling of the feed gas. After the catalyst had been loaded the spherical joints at the two ends of the reactor tube were connected to the respective gas lines and secured with the help of pinch clamps. A very small amount of silicon grease (Dow Corning) was applied to the joints to ensure leak free connections. Nitrogen gas (Linde, Ultra High Purity, 99.999%) was used as the inert carrier in the reactor. The nitrogen pressure was set at 10 psig from the pressure regulator on the

PAGE 42

35 gas cylinder and was never disturbed throughout the experiments. A needle valve was used to regulate the inert carrier flow while the downstream by-pass valve was kept at position 1 (reactant feed column by-pass) in the Figure 5. The gas flow rate was measured with the help of a soap bubble meter provided at the down stream end of the product gas line. Once the carrier flow had been adjusted the needle valve was not disturbed until all the catalysts had been processed. A specially designed reactant feed column, also shown in Figure 5, was used to feed cyclohexane vapor to the reactor. The bottom part of the 12 mm diameter pyrex column was filled with liquid cyclohexane at room temperature. The cyclohexane liquid level was brought up to a premarked level before starting the reaction with each catalyst sample. Nitrogen carrier gas was introduced in the feed column just above the liquid surface to sweep the cyclohexane vapor inside the feed column and exit at the top end of the reactor. This arrangement was used to avoid excessive entrainment of cyclohexane which might result if a bubble column were used. Before this design was adopted a bubble column was used to saturate the carrier gas with cyclohexane vapor. The result was a very high entrainment of cyclohexane in the feed gas. Every effort was made to avoid excessive entrainment or to remove the entrained liquid from the feed gas but with little success. The entrained liquid would settle in the feed lines and the concentration of the reactant would

PAGE 43

36 continue to increase and not stabilize even after several hours of operation. The entrainment problem was eliminated when the feed gas was introduced above the surface of the liquid reactant as described above. The feed column was immediately followed by a liquid trap which was equipped with a teflon stopcock at the bottom, to purge any liquid that may settle at the bottom of the trap. The upper portion of the liquid trap was used as a preheater which was heated by a coil of insulated nichrome wire wrapped on the out side of the glass tube. In the center of the preheater was a 2 mm thick glass fritt which was used to ensure proper heating of the feed stream The feed gas was then transmitted to the quartz reactor through a 1/8 inch polyethylene tubing which was insulated with fiber glass tape on the outside. The product line from the differential reactor was passed through a six port zero volume stainless steel gas sampling valve (Supelco, model 2-2915) mounted on the sample side of the gas chromatograph and then vented through a soap bubble meter into the fume hood. A 0.25 ml stainless steel sample loop (Supelco, model 2-2640) was used with the gas sampling valve to take samples of the product stream for analysis. A Hewlett Packard model 5790A series gas chromatograph equipped with a thermal conductivity detector was used for the analysis of the reaction products. The gas connections in the original chromatograph were altered to install a six

PAGE 44

37 port zero volume stainless steel gas sampling valve (Supelco, model 2-2915) on column A of the gas chromatogra ph which was used for the sample gas. The gas sampling valve was mounted right above the column oven and connected directly to the chromatographic packed column bypassing the Injection port on the column A. This was done in order to maintain the length of metal tubing between the gas sampling valve and the packed column because a long tubing could cause excessive peak broadening and thus introduce error. The reaction products were separated with the help of a 1/8 in. X 6 foot stainless steel column (Supelco) packed with di-n-decyl pthal ate on Chromosorb P (20 vttZ) (Maggiore et al., 1979). The column oven was programmed from 130 °C to 160 C at a rate of 4 °C/min followed by a 6.5 min isothermal operation at 160 °C. The detector temperature and the temperature of the injection port were kept at 250 °C. These conditions were chosen after many trial runs to get sharp separation of the product peaks at the fastest rate. A Hewlett Packard model 3790A reporting integrator was used to record the thermal conductivity cell response from the chromatograph. The integrator was attached to the chromatograph with the help of a remote starter cable in order to synchronize the run of the chromatographic oven with that of the integrator. Helium (Linde, Ultra High Purity, 99.999%) was used as the carrier gas in the chromatograph, and was further purified by passing through a molecular sieve trap (4°A, Davison) for the removal of water. The

PAGE 45

38 helium flow rate was kept at 20 ml/min through both the reference and the sample sides of the chromatograph. The nitrogen gas used as the inert carrier in the differential reactor was purified by passing through a bed of activated carbon (American Scientific Products) followed by a bed of 13x molecular sieve (American Scientific Products) to remove hydrocarbons and moisture respectively. The flow rate of the gas in the reactor was about 10 ml/min. Experimental Procedure The chromatograph was started in accordance with the instructions given in the operator's manual for the hewlett Packard 5790A series chromatograph. The helium supply pressure was set at 50 psig from the gas cylinder and the flow rate of the gas adjusted to 20 ml/min using the flow regulator valves provided on the front panel of the chromatograph and with the help of the soap bubble meter. After the carrier flow rate had stabilized the column oven was turned on and its temperature brought to 130 C mark. The detector temperature was then set at 250 C and the system allowed to stabilize for eight to ten hours. In addition to this at least one temperature programmed run was made on the chromatograph before processing any samples of the product gas to clean the packed columns. The detector current was turned on at least two hours before conducting any experiments to allow sufficient time for the system to stabilize. For overnight shut downs the detector sensitivity was either lowered or turned off to protect the detector elements from

PAGE 46

39 excessive wear and tear. The thermal conductivity detector -3 output was calibrated with the help of 1 x 10 ml injections of a standard mixture of cyclohexane and benzene. -3 Another 1 x 10 ml injection of deionized distilled water was used to calibrate the H^O peak. Each catalyst sample was dried for six to ten hours depending on whether it was powdered or in the form of pellets. Without this drying time excessive amount of H^O would be detected in the product stream. Visual observation determined that the moisture released from the catalyst, pores upon heating in the reactor tube, would settle in the uninsulated spherical quartz joint at the end of the quartz reactor and then continue to humidify the effluent gas stream at a very slow rate. Nevertheless, six hours of drying was sufficient to remove all the moisture and the drying was considered to be complete when no moisture was detected in the product gas. It must be mentioned that the drying was carried out at the reaction temperature and in an inert helium environment. Also, since the drying temperature was lower than the calcination temperature of 500 C, it is believed that this procedure did not effect other properties of the catalyst. It is believed that the only effect of drying was the removal of moisture from the surface of the catalyst. After drying, the carrier bypass valve was switched to position number 2 in the Figure 5. Thirty minutes were allowed for the reaction to stabilize before taking the

PAGE 47

40 first sample for analysis. It took 18.7 min for the chromatographic oven to complete one run and return to the ready state again. The following run was started as soon as the oven returned to the ready state. A total of four runs was made for each catalyst sample. Catalyst Preparation Five catalyst samples of moly-alumina catalyst with 0.1, 5, 10, 15, 25 wt%MoOper gram of catalyst were prepared by impregnating high surface area alumina (Norton, SA-6173; 1/16 in pellets) with an aqueous solution of ammonium molybdate (Fisher; A-674). Dei on i zed distilled water (4.4 ml/5 galumina) was used to to prepare the impregnating solution. The amount of ammonium molybdate used for a particular catalyst was calculated to give the desired loading of MoO^ on alumina. The actual amounts of the salt and water used are given in the appendix. The aqueous solution prepared was just enough to completely saturate the support material. Care was taken to ensure that very little excess solution remained in the crucible after all the solution had been added, and at the same time no alumina pellets were allowed to be left dry. The system was then allowed to stand for 22 hours followed by 24 hours of drying in the oven at 110 C. The dried samples were transferred to a high temperature furnace where the catalysts were calcined in air for 17 hours at a temperature of 500 °C. After calcination the samples were allowed to cool in air and then they were

PAGE 48

41 transferred to glass sample bottles for storage. The total weight of each sample prepared was 20 gm. Five catalyst samples of moly-silica catalyst with 0.1, 5, 10, 15, 20 wt%MoOloading per gram of silica were also prepared in the same way as the moly-alumina catalysts. High surface area silica catalyst support (Alfa, 80396) was impregnated with an aqueous solution of ammonium molybdate. Deionized distilled water (12.5 ml/5 gsilica) was used to prepare the impregnating solution. The actual amounts of the salt and water used for each catalyst are given in the appendix. Impregnation was allowed to continue for 24 hours. The catalyst seemed almost dry at the end of the impregnation period but was dried further for 24 hours in an oven at a temperature of 110 °C. Calcination in air was carried out for 18 hours in a furnace at 500 °C. The calcined sample was crushed to obtain a uniform powder and was stored in glass sample bottles. Experimental Results And Discussion Nitrogen BET was used to determine the surface area of each catalyst using the continuous flow sorptometer. The results indicate that molybdenum oxide is present in the form of a monolayer on the surface area of alumina for up to 15wt%Mo02 ^o^^i^S catalyst. That is to say that the dispersion of MoO^ is very close to unity in the case of the 5, 10, and 15 wtXMoO^ loading catalysts. The 25 wt%Mo02 loading moly-alumina catalyst is believed to carry MoO^ crystallites and its behavior is therefore different from the lower

PAGE 49

42 loading MoO^ loading moly-al umi na catalysts. In the case of the moly-silica catalysts the BET results suggest the formation of MoO-, crystallites for even the 5 wt%MoO^ loading catalyst. A detailed discussion of the BET results for the moly-al umi na and moly-silica catalysts follows. The BET surface areas of the moly-alumina catalysts are given in Table 3. Figure 6 shows the decrease of total catalyst surface area of different moly-alumina catalysts as the MoO^ loading is increased. However, if the total surface area of the individual catalysts is calculated per gram of alumina, it is seen that it remains constant for the 5, 10, and 15 wt%MoO^ catalysts, and then increases significantly for the 25 wtXMoO^ catalyst. This observation that the total surface area per gram of alumina for the 5, 10, and 15 wt%MoO^ loading moly-alumina catalysts remains constant, is consistent with published data (Liu and Weller, 1980). It conforms to the monolayer model for the moly-alumina catalysts which suggests that in the unreduced catalyst molybdenum oxide is present in the form of an epitaxial monolayer on the surface of alumina. The monolayer model for the unreduced moly-alumina catalyst is a widely accepted model in the published literature (Massoth, 1973). Massoth, for instance, has used butene chemi sorp ti on at 100 C on freshly prepared 10 wt^MoO^ loading moly-alumina catalyst to show a butene to molybdena ratio of 0.63 which compares well with his theoretical value of 0.64 obtained by assuming that all the molybdena were available on the surface for

PAGE 50

43 Table 3. Nitrogen BET surface area of moly-al umi na catalyst. Area Area 2 2 wt% MoO^ m /gcat m /galumina 0.1

PAGE 51

44 205 185. 185 175 165 15 5(| 145 • Area per gram catalyst O Area per gram alumina 10 15 wt%MoO. 20 25 Figure 5. BET surface area of moly-alumina catalyst .

PAGE 52

45 adsorption. Massoth has also observed that only the 25 wtXMoO^ loading catalyst gave a sharp XRD peak for MoO^, suggesting that crystalline moly is present only in the higher loading moly-al umina catalysts. Stencel et al. (1983) have studied M0O2/AI2O2 with the help of Raman Spectroscopy and XRD. Their results also show that crystalline MoO^ is present only in the higher loading moly-al umi na catalysts. Therefore we conclude that the increase in the BET surface area (per gram of alumina) for the 25 wt%MoOloading molyalumina catalyst is due to the presence of crystalline moly which contributes to the total surface area of the catalyst. For the catalysts with up to 15 wt%MoOloading the molybdenum oxide appears to form an epitaxial monolayer on the surface of alumina. The BET results for the moly-silica catalyst are given in Table 4. It is seen that the total surface area of the moly-silica catalysts decreases with increased MoO^ loading. This can be explained by viewing the structure of the molysilica catalyst which is quite different from that of the moly-alumina catalyst. No substantial evidence in the literature was found which may suggest the formation of a monolayer on the surface of silica. In fact, Garcia Fierro et al. (1980) have shown the presence of MoO^ crystallites on the surface of silica for a 13 wt%MoOloading molysilica catalyst with the help of a SEM micrograph. Even though the total surface area of the moly-silica catalyst remained constant for Garcia Fierro et al., in all the

PAGE 53

46 Table 4. Nitrogen BET surface area of moly-silica catalyst. Area Area 2 2 wt% MoO^ m /gcat m /gsilica 0.1

PAGE 54

47 300 250 CM S 200 i. 150 100 • Area per gram catalyst O Area per gram silica 10 wt%MoO. 15 20 Figure 7. BET area of moly-silica catalyst

PAGE 55

48 catalysts the total surface area of the catalyst was much less than the area that should have been contributed by the amount of silica support present in the catalyst samples. In addition to this the pore volume of their catalyst also 3 decreased from 1.34 cm /gm for silica to 1.03, 0.86, and 3 0.75 cm /gcat for the catalysts containing 4.8, 8.1 and 13.0 wt^MoO^* respectively. Thus it was concluded that the MoOcrystallites block the pores in silica and this results in reduced surface area of the catalyst as the MoO^ loading is increased. This is in contrast to the behavior exhibited by the 25 wt% moly-alumina catalyst where the MoO^ crystallites contributed to the total surface area of the catalyst. However, it may be recalled that the alumina used in the moly-alumina catalyst was in the form of pellets and its pore size distribution may be different from the pore size distribution of the silica used in the moly-silica catalyst. Thus, while the moly crystallites may block the pores of silica they may not block the pores of the alumina used in the moly-alumina catalyst. The results of nitrous oxide selective phy si sorp ti on indicate that nitrous oxide is not the ideal adsorbate for the moly-alumina catalyst if the absolute value of the fractional surface area is of interest. Also it is observed that the values of the proportionality constants b. in Eq. (10) cannot be unity for the nitrous oxide selective physisorption on the moly-alumina catalyst. Therefore, nitrous oxide selective phy si sorpti on cannot be used to determine

PAGE 56

49 the absolute value of the fractional catalyst surface area of moly in the moly-alumina catalysts. Nevertheless, as will be seen latter, ratios of the fractional surface areas of different moly-alumina catalysts with up to 15 vit%l^oO^ loading can still be evaluated with the help of nitrous oxide selective phy sisorpti on data. In the case of the moly-silica catalyst, it was found that nitrous oxide selective phy sisorpti on does not yield any useful information regarding the fractional catalyst surface area. A detailed discussion of these results follows. The results of nitrous oxide selective physi sorpti on at -78 °C on the M0O3/AI2O3 catalyst, and the Mo02/Si02 catalyst are shown in Figures 8 and 9, respectively. Each data point shown in these two figures represents the average of the two experimental values obtained for each catalyst sample being considered. The error analysis indicates that the values of the packing factors may be ±2% in error (Appendix F). In light of this fact the differences between the Yt values of the 5, 10, and 15 wt^MoO^ loading molyalumina catalysts (about 2% each) may not seem to be significant. However, it was noted that the maximum scatter in the experimental data for any individual catalyst never exceeded 1.6% and in most cases was less than 1.0%. Also, because the difference between the minimum and the maximum y^ values is much greater than 2%, we believe that the differences shown are indeed significant and the trend shown by the data is real.

PAGE 57

50 105 .100 i. o +-> (J fO <4cn o Q. >09 5 090. .085 030 10 wt%Mo0 15 20 25 Figure 8 Nitrous oxide selective physi sorption on moly-alumina catalysts.

PAGE 58

51 .100 o o IB CD C •r— o Q. ?.094 .092 .090 5 10 15 20 wt%Mo03 Figure 9. Nitrous oxide selective physi sorpti on on moly-silica catalyst.

PAGE 59

52 In Figure 8 it is seen that the nitrous oxide packing factor for the moly-alumina catalyst increases with increased MoOloading with up to 15 wtXMoO^ loading, and then decreases sharply for the 25 wt^MoO^ loading catalyst. This behavior can be explained by viewing the different structure of the higher loading catalyst. As it was discussed earlier, the molybdenum oxide in the 25 wtXMoO^ catalyst appears to be present in the form of MoO^ crystal2lites along with the one present as Al^C^oO. )^ (Massoth, 1973). Since the packing factor for the pure MoO^ is much less than the packing factor for pure alumina the MoO^ crystallites in this catalyst may force the value of the packing factor for the 25 wt%MoOloading catalyst to move towards' the pure MoO^ value. Thus the packing factor for the 25 wt%Mo0sub3 loading catalyst is much lower than those of the lower loading moly-alumina catalysts. Because of this complexity, the theory of selective phy si sorpti on cannot be rigorously applied to the 25 wt%MoO^ catalyst. For the lower loading moly-alumina catalysts, it is clear that the packing factors for the dispersed catalysts are higher than the packing factors for the pure components. Therefore the b. in Eq. (10) cannot be unity for selective phy sisorpti on of nitrous oxide on the moly-alumina catalyst. In order to get any meaningful results from these data we assume that the catalyst with 0.1 wt^MoO^ loading has negligible fractional catalyst surface area. Considering that the total surface area of the individual catalysts is of the

PAGE 60

53 order of 170 m /gcat, this assumption does not seem to be unrealistic. In that case then, the 0.1 wt%MoO, catalyst may be considered as the pure support with bp equal to unity. Therefore, from Eq. (10) we get R.x.(T) = b,Y,(T) 'n IM (13) and R2X2(T) = Y2 (T) (14) Substituting these expressions in Eq. (5) for catalysts j and k t J '-'^,h'1 ['1^1 TTry Y2(T) m — -] ] (15) and t k '-'slh ^72(1) ^ 1 CbiY^ TTT TTnT (16) Divide Eq. (15) by Eq. (16) to get ^1 (17)

PAGE 61

54 which can be used to calculate the ratios of the fractional catalyst surface areas of different catalysts. The results of this calculation for the moly-al umi na catalysts are given in Table 5. Thus even in the case where b. are not unity it is possible to at least evaluate the ratios of the fractional surface areas of different wt% loading catalysts using a catalyst with negligible wt% loading as pure support. Theoretical values of fractional surface area ratios of different moly-al umi na catalysts, calculated by assuming a 22complete monolayer formation of AlpC^oO. )_ and a MoOcross section of 25 °A (Massoth, 1973), are also given in Table 5. It is seen that the theoretical values do not compare well with the experimental values. Also, if indeed the monolayer model for the moly-alumina catalyst were valid, and in light of the fact that the analysis of the BET results does point towards the formation of an epitaxial monolayer of molybdenum on the surface of alumina, the results of the theoretical calculation should not be very far from the actual values of the fractional surface area ratios. On the other hand, it must be born in mind that the theoretical calculations only represent an upper limit for the fractional surface area ratios and the calculated values are only hypothetical in nature. In any event, it is clear that the values calculated from the nitrous oxide selective physi sorption are much lower than the theoretical values.

PAGE 62

55 M

PAGE 63

56 Since little is known about the s toi chi ome try of the interactions between the selectively physisorbed gas and the catalyst surface, we are unable to explain as to why the nitrous oxide selective phy si sorpti on results are much lower than the theoretically calculated values of the fractional surface area ratios. Nevertheless, it is interesting to note that the results of the two calculations follow the same trend. It may also be pointed out that the fractional surface area of a 15 wt%MoO^ moly-alumina catalyst obtained by Parekh and Weller (1977), using low temperature oxygen chemisorption, was also about one fourth of the value that may be obtained if a monolayer model were adopted. From this discussion it is clear that more work needs to be done in order to understand and explain the reason for the disagreement between the results of the nitrous oxide selective physisorption and the theoretical results obtained by assuming molybdenum monolayer on the surface of alumina. Since there is no other independent method available to measure the fractional surface area of the oxide catalysts, an absolute verification of the nitrous oxide selective phy si sorpti on method was not possible. Therefore, in order to make a qualitative comparison, it was decided to compare the cyclohexane dehydrogenati on activity of the oxide catalysts with the nitrous oxide packing factors obtained from the selective physi sorption of nitrous oxide. The nitrous oxide selective phy sisorpti on results for the moly-silica system do not yield any useful information

PAGE 64

57 regarding the fractional surface area of the catalyst. At best the packing factors for the moly-silica catalysts of different wt%MoO^ loadings show a decreasing trend up to 15 wt%MoO^ loading catalyst, but due to the more complex nature of this catalyst a better explanation cannot be offered at present. The sudden increase in the value of the packing factor for the 20 wtXMoO, loading moly-silica catalyst is not understood. The cyclohexane dehydrogenati on experiments were carried out in a differential reactor and the results for the moly-alumina catalyst system and the moly-silica catalyst system are shown in the Figures 10 and 11, respectively. The data shown in these two figures represents the run number two for each catalyst being considered. It may be recalled that four samples of the product gas were analyzed for each catalyst sample. Since a differential reactor was being used it is important to compare the activity of each catalyst after it has been exposed to the same reaction conditions for the same duration of time. The error analysis indicates that the calculated values of the catalyst activity may be +5% in error. The rate of reaction was calculated from '•c = ^ ^ A-W ^ (18) where F is the molar feed flow rate and AX and AW are the differential conversion and the weight of the catalyst.

PAGE 65

58 respectively. Cyclohexane, benzene and water were the only components detected in the product stream. Since the reactor was being operated in differential mode and the amount of coking did not increase significantly with the increased catalyst loading, it was neglected in the calculation of the rate of reaction. The conversion was calculated by assuming the reaction S"i2 -' ^eh ^ ^h (19) only. The small amount of the HgO present in the product stream may be attributed to the water attached to the Al2( MoO^ )^ species which could have been released upon cyclohexane chemisorpti on on the surface of the catalyst. It merits mention that the concentration of the H^O reduced drastically in the first two runs and only trace amount was detected in the subsequent runs, while the concentration of cyclohexane and benzene remained constant in all the runs. This observation fortifies our contention that the H„0 present in the product stream does not represent a product but is merely the water chemisorbed on the catalyst which is released when cyclohexane chemi sorpti on starts. Figure 10 shows that no reaction was observed on pure alumina or the 0.1 wt^MoO^ catalyst. After that the rate of reaction increased steadily from 5 wt%Mo03 to 15 wt%Mo03 catalyst, and then the rate of increase of the reaction rate slowed down towards the 25 wtlMoOcatalyst. It is believed

PAGE 66

59 1600

PAGE 67

60 ^o 400 a; ea on 10 15 wt% MoOo 20 25 Fi gure 11 Rate of cyclohexane dehydrogenati on on moly-silica catalyst.

PAGE 68

61 that this change in the rate of increase of the reaction rate is primarily due to the change in the nature of the catalyst surface as the loading increases from 15 to 25 wtXMoO^. It has been discussed earlier that MoO^ crystallites may be present on the surface of the 25 wt%MoO-> catalyst. The present observation further strengthens our belief that the 25 wtXMoO^ catalyst is different in nature from the lower loading catalysts and, therefore, must be treated separately. The correlation between the nitrous oxide packing factors and the cyclohexane dehydrogenati on activity of the moly-alumina catalysts is shown in the Figure 12. It is seen that the activity increases with the increase in the value of the packing factor. This, however, is not the case for the 25 wtXMoO^ catalyst but this catalyst has already been shown to be an exception. Again, in spite of the relatively small differences in the individual values for the 5, 10, and 15 wt^MoO, catalysts (about 2% each), the correlation is believed to be significant because of the good reproducibility experienced in the selective phy si sorp ti on experiments and the overall trend of the data. Thus, the correlation shown in the Figure 12 is believed to present a valid partial verification of the nitrous oxide selective physi sorpti on results. In the case of the moly-silica catalyst also, the cyclohexane dehydrogenati on activity is seen to increase with increased MoO^ loading (Figure 11). Also, as in the

PAGE 69

62 O 1600

PAGE 70

63 case the moly-al umi na catalyst, no reaction was observed on pure silica or the 0.1 wtXMoO^ catalyst. The rate of reaction increased steadily with increase in the MoO^ loading and no shift in behavior was observed for the higher loading catalyst. This observation along with the BET results shown in the Figure 6 leads us to believe that all moly-silica catalysts may be similar in nature in the sense that all crystalline MoOmay be present even in the lower loading catalysts. No useful correlation between the activity of cyclohexane dehydrogenati on and the nitrous oxide selective phy sisorpti on results could be obtained for the moly-silica catalysts. Therefore we conclude that nitrous oxide selective phy si sorpti on does not provide any useful information regarding the fractional catalyst surface area of the molysilica catalyst system. In order to further study the moly-alumina catalyst surface, the 10 wt%Mo02 catalyst was analyzed using X-ray photoelectron spectroscopy. The survey scan for the 10 wt%MoOcatalyst is shown in the Figure 13. The positioning of the individual peaks on the binding energy scale can be adjusted for the experimental work function by taking the Al 2p peak as the reference peak. According to the data given in the handbook of X-ray photoelectron spectroscopy (Wagner et al., 1979), the binding energy of the Al 2p peak for AKO^ is 74.1-74.3 eV, and the binding energy of the Al 2p 2peak for Al2(MoO. ) is 74.4 eV. Since only a fraction of 2the total aluminum may be present as Alp(MoO. )^, the Al 2p

PAGE 71

64 peak is moved to 74.2 eV which is the average value of the binding energy for AKO^. This gives an experimental work function of 4.85 eV. Using this experimental work function value the binding energies of the Mo 3d, Is, and C Is peaks come out to be 232.85 eV, 531.4 eV, and 278.9 eV, respectively. It is interesting to note that the binding energy of the Mo 3d peak (232.85 eV) is very close to the the binding energy of the Mo 3d peak for Al2(MoO^^")3 (232.8 eV). Also, since the binding energies for MoO^ and MoOp are less than 232.7 eV, we may conclude that Mo is present on 2the catalyst surface as Al2(MoO. )-. A high resolution scan of the C Is peak is shown in the Figure 14. The binding energy of the C Is peak suggests that the carbon is present in the form of a carbide. A definitive source of the surface carbon is not known. However, the carbon may partially be present due to carbon dioxide and carbon monoxide present in the atmosphere which may have chemisorbed on the catalyst surface upon exposure to the air. It is quite likely that during calcination the chemisorbed carbon oxides left some graphitic carbon on the catalyst surface which in turn transformed into a carbide. The surface composition of the 10 wt%MoO^ moly-alumina catalyst obtained from the X-ray photoel ectron spectroscopy data is given in Table 6. It is seen that carbon represents 3.37 mass% of the total catalyst surface. However, it is not clear whether the carbon is present on the catalyst surface in the form of a thin monolayer or in the form of large

PAGE 72

65 o o
PAGE 73

66 o (NJ o

PAGE 74

67 (O 1

PAGE 75

68 crystallites. Also it is not known as to how the presence of this carbon may effect the nitrous oxide selective physi sorption on the catalyst because similar carbon may be present on all the catalyst samples as well as the two pure components. Therefore, we recommend a thorough investigation of these aspects when future work is undertaken. It will be important to identify, with certainty, the source and form of the surface carbon so that the effects of carbon on the method of selective phy si sorp ti on can be evaluated. Identification of the source may also aid in avoiding the carbon deposit if that is deemed essential for the method of selective phy si sorpti on to be effective. Summary The method of selective phy si sorp ti on for measuring the fractional catalyst surface area of dispersed catalysts has been applied to the oxide catalysts with moderate success. Five moly-alumina catalysts of different MoO^ loading and five moly-silica catalysts of different MoO^ loading were prepared by impregnating high surface area alumina and high surface area silica respectively, with aqueous solutions of ammonium molybdate. The total surface area of each catalyst sample was determined with the help of nitrogen BET method. The results of the BET study indicate the formation of a molybdenum oxide monolayer on the surface of alumina for the lower MoO^ loading catalysts. That is to say that the dispersion of molybdenum oxide on the surface of alumina is unity for the 5, 10, and 15 wtXMoO^ catalysts, and is less

PAGE 76

69 than unity for the 25 wt%MoOcatalyst. This conclusion was drawn from the observation that the total surface area of the 5, 10, and 15 wt%MoOloading moly-al umi na catalysts remains constant when it is calculated per gram of alumina present in the catalyst. This behavior and the subsequent conclusion is consistent with the view of most other researchers in the literature who have studied the molyalumina system. The increase in the BET surface area of the 25 wt%MoOmoly alumina catalyst is attributed to the presence of MoOcrystallites on the surface of this catalyst. The MoO^ crystallites have been confirmed to exist on the higher MoO^ loading moly-alumina catalyst, with the help of XRD and Raman spectroscopy. The nitrous oxide selective phy si sorp ti on was applied to the moly-alumina catalyst. It was observed that the nitrous oxide packing factors for the moly-alumina catalysts increased with increased MoO^ loading for up to 15 wt^MoO^ loading catalyst, and then decreased sharply for the 25 wt%MoO^ loading moly-alumina catalyst. Also it was observed that the nitrous oxide packing factors for the moly-alumina catalysts with up to 15 wt%Mo02 loading are greater than the nitrous oxide packing factors for the pure iMoO^ and pure alumina. Thus it was clear that nitrous oxide was not the best suited adsorbate for the moly-alumina catalyst if the absolute value of the fractional catalyst surface area is of interest. Nevertheless, ratios of the fractional catalyst surface areas of different catalysts with different wtXHoO^

PAGE 77

70 loading can be computed if the 0.1 wtXMoO^ loading catalyst were considered as the pure component. The underlying assumption here was that the 0.1 wf^MoO^ loading molyalumina catalyst has negligible molybdenum oxide surface area and therefore could be used as a reference. Thus the fractional surface area ratios were computed to be 1.3, 1.6, and 1.2 for 10 wt% to 5 wt%, 15 wt% to 5 wt%, and 15 wt% to 10 wt % MoO^ loading moly-alumina catalysts, respectively. Since the surface characteristics of the 25 \^t%'AoO^ loading moly-alumina catalyst are very different from the rest of the catalysts it could not be included in the above calculation. The fractional surface area ratios mentioned above are much lower compared to the theoretical values calculated from the BET results assuming a monolayer of molybdenum oxide on the surface of alumina. However, the results of the two calculations follow the same trend. The results of nitrous oxide selective physi sorption for the moly-alumina catalyst system were substantiated by a correlation between the selective physi sorp ti on data and the cyclohexane dehydrogenati on activity for the moly-alumina catalysts with up to 15 wt^MoO^ loading. The 25 wfb'loO^ catalyst was not included in the correlation because of its different surface characteristics. The correlation is believed to be significant in spite of the relatively small differences in the individual packing factor values because the scatter in the scatter in the experimental values was even smaller. The overall trend of the data also strengthens

PAGE 78

71 this belief. The moly-al unii na catalyst was also analyzed using X-ray photoel ec tron spectroscopy to further study the surface of this catalyst. The results indicate that a substantial amount of carbon is present on the surface of the catalyst in the form of a carbide. The exact source and form of this surface carbon are not yet known with certainty. Also it is not known how this carbon may effect the selective phy si sorpti on method. We recommend a more thorough study of this phenomenon in order to develop the method of selective phy si sorpti on further. In contrast to the moly-alumina system the total surface area of the moly-silica catalysts decreased with increased MoO^ loading. This was so even when the BET surface area of the catalysts was calculated per gram of silica present in the catalysts. This observation and the results of other researchers led us to believe that MoO^ crystallites are present on the moly-silica catalyst even when the catalyst loading is relatively low. The MoOcrystallites block the pores in the silica support and, therefore, the total surface area of the catalysts decreases with increased MoO, loading. The nitrous oxide selective phy si sorp ti on was not successful in determining the fractional catalyst surface area of the moly silica catalysts. At best a decreasing value of the nitrous oxide packing factor with increased MoO^ loading was observed for the moly-silica catalysts with up to 15 wt%MoOT loading. This trend reversed for the 20 wt%Mo02 loading catalyst and the reason for this behavior is

PAGE 79

72 not understood. No correlation could be obtained between the selective phy si sorpti on data and the cyclohexane dehydrogenation activity for the moly-silica catalyst system.

PAGE 80

CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS The method of selective phy si sorption for measuring the fractional catalyst surface area of dispersed catalysts as developed by Miller and Lee (1984) has been modified to achieve increased sensitivity towards changes in the fractional surface areas of catalysts with different loadings of the dispersed material. It has been shown that the use of packing factors (y) in place of fractional coverage (9) affords a better sensitivity towards measuring the fractional surface area of any catalyst. The theory has been accordingly modified to incorporate packing factors instead of fractional coverage. A few modifications have also been suggested in the experimental method used for obtaining the adsorption isobars of the selectively physisorbed gas. The new method incorporates the use of more than one constant temperature cold baths to obtain the adsorption isobars and thus eliminates the need for the pseudo steady state assumption that has been used by Miller and Lee (1984). Thus a more dependable adsorption isobar is obtained in this manner. One important objective of this research was to demonstrate the effectiveness of the method of selective physi sorption in measuring the fractional surface area of 73

PAGE 81

74 supported catalyst, and establish its credibility by using it to determine the surface area of a standard catalyst system. A standard catalyst system would be a one which would provide an already established alternate method of determining its fractional surface area. Platinum dispersed on silica was chosen because the fractional surface area of dispersed platinum can be determined very effectively with the help of hydrogen chemi sorpti on . Four samples of different wt%Pt loading on silica were prepared by impregnating amorphous silica with an aqueous solution of chl oropl atani c acid. Both carbon dioxide and nitrous oxide were used to independently determine the fractional surface area of the dispersed platinum in the four catalysts. The results indicate that carbon dioxide is not a suitable adsorbate for measuring the fractional catalyst surface area of platinum in the platinum silica catalyst, although it may provide valuable insight into the said system. Nitrous oxide, on the other hand, was found to be a good adsorbate to be used for the selective phy si sorp ti on on the platinum silica catalyst. This observation also demonstrated that our earlier contention, that directional pol ari zabi 1 i ty data can be used to select an adsorbate for selective phy si sorp ti on , was correct. The values of the fractional surface area of platinum dispersed on silica determined from the selective physisorption of nitrous oxide were found to be in agreement with the values obtained from the hydrogen chemi sorp ti on experiments.

PAGE 82

75 It was shown that nitrous oxide selective phy si sorp ti on is sensitive to even small changes in the platinum surface area as present in the supported catalyst. In order to advance the method of selective phy si sorption to metal base oxide catalysts, nitrous oxide selective phy si sorpti on was attempted on two i.ndustrial ly important oxide catalysts, namely moly-alumina and moly-silica catalysts. Five samples of each catalyst system were prepared by impregnating high surface area alumina and high surface area silica with aqueous solutions of ammonium molybdate. The total surface area of each catalyst was determined with the nitrogen BET method. The results indicate that the total surface area of up to 15 wt%MoO^ loading moly-alumina catalysts remained constant when calculated per gram of the alumina present in the catalyst, and the total surface area per gram of alumina of the 25 wt%MoO^ loading moly-alumina catalyst increased significantly. This observation, along with the data and view of other researchers in the published literature, leads us to the conclusion that moly oxide forms a epitaxial monolayer on the surface of alumina in the moly-alumina catalyst, with up to 15 wtXMoO^ loading, while crystalline MoO . . ,. ^ ^. • ^ + ,• „ ^' -^ T " ^ believed to oe present in the 25 wt%MoO, loading moly-alunina catalyst along with the moly oxide monolayer. Nitrous oxide selective phy si sorp ti on was only partially successful when applied to the moly-alumina catalyst. This is so, because the packing factors for the catalysts

PAGE 83

76 with up to 15 wfSMoO^ loading were greater in value than the packing factors for the pure MoO^ and pure alumina. This clearly showed that b. for this system could not be unity. Therefore absolute values of the fractional catalyst surface areas of these catalysts could not be determined. Nevertheless, ratios of the fractional surface areas of different moly-alumina catalysts can still be calculated by treating the 0.1 wt%MoO^ loading catalyst as the pure support. The underlying assumption here is that the fractional catalyst surface area of the 0.1 wtXMoO^ catalyst is negligible compared to the total surface area of the catalyst. The 25 wt%MoO^ catalyst was treated as an exception, and was not included in the calculation being discussed here, because it is believed to carry crystalline MoO, on its surface which makes the nature of its surface different from the nature of the surface of the catalysts with lower catalyst loadings. The results of this calculation do not compare well with the results of theoretical calculations based on the monolayer model for the moly-alumina catalysts. An absolute verification of the nitrous oxide selective phy si sorption results was not possible because no other independent method was available to measure the fractional surface area of the oxide catalyst. Instead a partial verification of results was made using a correlation between the cyclohexane dehydrogenati on activity and the nitrous oxide packing factors for the moly-alumina catalysts. The correlation is believed to be significant, in spite of the

PAGE 84

77 relatively small differences between the individual packing factor values, because the scatter in the experimental values was even smaller. The overall trend of the data also suggested that the correlation was indeed significant and valid. The BET results for the moly-silica catalyst system indicate the presence of MoO-. crystallites on the surface of silica which results in decreased total surface area of the catalyst with increased MoO^ loading. Nitrous oxide selective phy si sorpti on on the moly-silica catalysts was not successful, and it was determined that nitrous oxide is unsuitable for use as an adsorbate for the moly-silica catalyst. No useful correlation could be obtained between the results of nitrous oxide selective phy si sorpti on and the cyclohexane dehydrogena ti on activity of the moly-silica catalysts. The 10 wt%MoO^ moly-alumina catalyst was analyzed using X-ray photoel ectron spectroscopy to study the surface further. It was observed that in addition to the oxygen, molybdenum, and aluminum present on the surface of the catalyst there was a substantial amount of carbon also present on the surface. The exact source and form of the surface carbon is not yet understood, but the carbon is believed to come from carbon monoxide and carbon dioxide present in the air. Also it is not known how this carbon may effect the selective phy si sorpti on method. Therefore, it is recommended that such surface analysis may be included in future work in order to develop the theory of selective

PAGE 85

78 phy si sorp ti on method further. It will also be important to correctly identify the source of the carbon present on the surface of the catalyst so that it may be avoided if it adversely effects the method of selective phy si sorpti on for measuring fractional catalyst surface area. It is the author's understanding that more v/ork needs to be done to better understand the nature of the interactions between the selectively physisorbed gas and the solid surface. Infra red spectroscopy may be able to offer some insight into the state of the selectively physisorbed gases which will help in establishing a better and more effective criterion for the selection of a suitable adsorbate. Since nitrous oxide selective phy sisorpti on was unable to determine the absolute values of the fractional surface areas of the moly-alumina catalysts and nothing for the moly-silica catalysts, we recommend more work in this area. A better adsorbate needs to be selected to obtain more conclusive results. The author suggests acetylene as a suitable candidate for use with the moly-alumina and moly-silica catalyst systems . An important application of the method selective physisorption may be in studying catalyst deactivation by coking. Preliminary results indicate a large difference between the nitrous oxide packing factors for pure nickel and pure graphite powders. The method may also be helpful in differentiating between the catalyst deactivation caused by

PAGE 86

79 coking and the catalyst deactivation caused by sulfur poisoning when the tv/o occur simultaneously.

PAGE 87

APPENDIX A PREPARATION OF OXIDE CATALYSTS Table 7. Preparation of moly-al um1 na catalyst, VJt% M0O3

PAGE 88

81 Table 8. Preparation of moTy-silica catalyst, Ammoni um wt% MoO-, Silica Molybdate Water (gm) (gn) (ml) 0.1

PAGE 89

APPENDIX B NITROUS OXIDE SELECTIVE P H YSI SOR PTION DATA FOR PLATINUM-SILICA CATALYST Table 9 Experimental conditions for nitrous oxide selective phy si sorpti on on pi a ti numsi 1 1 ca catalyst. Helium pressure (in H-O) 25.1 Nitrous oxide pressure (in H^O) 1.8 Nitrous oxide Flow meter reading 30.0 Attenuation of desorption peak xl Attenuation of calibration peak x4 Volume of calibration gas (cm~^) 0.179 Area of calibration peak 99.3 Heater voltage (V) 15^ Total flow rate (cm /min) 33.55 Helium flow rate (cm^/min) 30.185 Concentration chart speed (in/min) 2 Temperature chart speed (in/min) 0.5 82

PAGE 90

83 Table 10. Experimental data of nitrous oxide selective phy si sorp ti on on platinum-silica catalyst. Sampl e Run K-Type Peak Thermocouple IJ eight Area Reading (mV) (gm) 1.0

PAGE 91

84 Table 10--continued.

PAGE 92

Table 10--continued. 85 Sampl e Run K-Type Peak Thermocouple Weight Area Reading (mV) (gm) Pt 7 8 9 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 222.3 225.9 226.9 232.1 46.1 43.4 72.8 68.4 75.7 74.7 76.0 71.8 72.4 73.6 84.5 85.4 85.9 86.4 2.85 2.85 2.88 2.91 1.73 1.70 2.47 2.41 2.67 2.63 2.60 2.56 2.51 2.57 2.82 2.85 2.86 2.90 0.39557

PAGE 93

APPENDIX C NITROUS OXIDE SELECTIVE PH YS I SORPT I ON DATA FOR THE OXIDE CATALYSTS Table 11. Experimental conditions. Helium pressure 4. Nitrousoxidepressure 3,5 Attenuation of desorption peak xl6 Attenuation of calibration peak x4 Volume of calibration peak (cm ) 0.25 Area of calibration peak 38.5 Concentration chart speed (in/min) 4 Temperature chart speed (in/min) 4 Heater voltage (V) 8.7 Detector current (mA) 80 Detector temperature (°C) 60 Total flow rate (ml/min) 38.61 Helium flow rate (ml/min) 35.62 86

PAGE 94

87 Table 12. Experimental data for nitrous oxide selective phy si sorption on moly-aluinina catalyst. Sample Run Peak Temperature Weight Area (°C) (gm) 5 wt%

PAGE 95

88 Table 13. Experimental conditions for nitrous oxide selective phy si sorption on moly-silica catalyst. Heliumpressure 3.4 Nitrousoxidepressure 3.0 Attenuation of desorption peak x32 Attenuation of calibration peak x4 Volume of calibration peak (cm ) 0.25 Area of calibration peak 52.4 Concentration chart speed (in/min) 4 Temperature chart speed (in/min) 4 Heater voltage (V) 8.7 Detector current (mA) 80 Detector temperature (°C) 60 Total flow rate (ml/min) 38.61 Helium flow rate (ml/mi n) 35.62

PAGE 96

89 Table 14. Experimental data for nitrous oxide selective phy si sorption on moly-silica catalyst. Sample Run Peak Temperature Weight Area (°C) (gm) M003

PAGE 97

NITROGEN APPENDIX D PHYSISORPTION RAW DATA

PAGE 98

91 o> o o CVJ CO X fO r-l O CM CVJ CM t-H CO CM X X o 00 CTI IT) CM CO CM 00 IT) o io +-> n3 CM CM CM X CO CM 00 CO 00 "Sio> X 00 00 in o i. 3 s. o U) ir> o CM at
PAGE 99

92 <0 o T3 it-> c o rrs +-> T3 00 CO oo

PAGE 100

93 CM o X ro o 00 O X VO CO 00 00 IT) CM O CM vo CM CM CM O X 00 ^£5 1/1 <0 o CM o I E 3 o ID o CM o cv CM CM X rH CM CM X CO CTl X O CM CM CM 1 — CM »-H X o cr> (T3 n3 o CO o J2 <0

PAGE 101

94 CM CO CM r-H X CvJ >< CM CO n CM CM Lf) CM CO o 3 o CM cri X CO CM o IT) o X Oi CM CO ro CO 00 CTl X CO CM CO Lf) o un CM .-H X to +J o CO 00
PAGE 102

95 in CM ID O o «3 CM CM X aX U3 +-> m >. +-> <0 o o CM lO o CM VO O CM o IX) CM o X X o 4-> IT3 CNJ

PAGE 103

96 cv 1—) o CM X '-I 9\ X >> 03 O I/) I E =3 CM 00 ro ID

PAGE 104

97 •o o (B 0) s~ +-> s_ a. so <0 •a CM <0 ro CM

PAGE 105

98 CO o in

PAGE 106

99 o CO

PAGE 107

100 LO CM

PAGE 108

101 CM LT) CTl ^ CO CM O C CT.

PAGE 109

102

PAGE 110

103 »-> ro O (T3 B I &5 3

PAGE 111

104 CTi m CX3 o O CO
PAGE 112

105
PAGE 113

106

PAGE 114

107

PAGE 115

108 +-> 00 O I C\J 0^ CNJ CM O n
PAGE 116

109 to O CO CTl

PAGE 117

APPENDIX E DIFFERENTIAL REACTOR DATA Composition of the calibration sample -3 Injection size = 10 ml. Area of the C^H^^ peak = 7997900. Area of the CgHg peak = 989530. 90% CgH^2> 10"^ CgHg 110

PAGE 118

Ill Table 34. Raw data for cyclohexane dehydrogena ti on on moly-alumina catalyst. Area of Peak Sample Run H2O ^6^12 S"6 Weight 0.1 wt% 5 wt% 10 wt% 15 wt% 25 wt% 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 2772100 2765800 2768700 2736500 2733800 2726200 2729900 2695400 2677400 2683200 2688200 2700400 2696300 2687400 2701300 2639800 2657100 2662000 2667600 8283 12322 13337 10308 10238 8095 24239 16161 12767 11198 24510 14926 12976 9907 63255 21910 16954 15420 228700 225090 223450 247950 247040 245290 244130 231510 222650 220640 216260 212250 213440 216220 214040 214630 223140 223080 219990 3720 4007 4301 4698 12897 13362 20040 20319 17069 20752 22736 22874 20837 26947 28707 29479 1.0008 1.0048 1.0121 0.6394 0.6828 Feed flow rate = 10 ml/56.8 sec.

PAGE 119

112 Table 35. Raw data for cyclohexane dehydrogenation on moly-silica catalyst. Sample Run N, Area of Peak H^O ^6^12 ^6^6 Weight 5 wt% 10 wt% 15 wt% 20 wt% 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 2750900 2749600 2750100 2742600 2772200 2756900 2750300 2750100 2766500 2753300 2748600 2747000 2694300 2750400 2752700 2753100 10698 7183 5955 13018 9118 7533 7202 21658 8665 6872 6297 123530 12218 8944 7879 279670 279070 274340 270570 273580 277090 275910 269320 253360 263380 257400 256490 223440 237020 238760 237230 10 28 378 722 568 2968 1760 1503 1434 10703 5719 4566 3315 13960 8131 6063 5144 0.6022 0.5015 0.6039 0.5960 Feed flov; rate = 10 ml/57.94 sec

PAGE 120

APPENDIX F ERROR ANALYSIS OF THE PACKING FACTOR VALUES Error contributed by: 1. The thermocouple cold junction = +0.2% 2. The K-type thermocouple wire = +0.4% 3. T.C. Detector plus integrator = +1.0% 4. Temperature recorder = +0.5% Total error = +2.1% 113

PAGE 121

LIST OF REFERENCES Adler, S.F., and J.J. Kearney, J. Phys. Chem., 64, 208 (1960). Anderson, R., "Experimental Methods in Catalytic Research," Academic Press, New York (1968). Bartholomew, C.H., and M. Boudart, J. Catal., 29, 278 (1973). Brunauer, S., "The Adsorption of Gases and Vapors-Physical Adsorption," Princeton University, Princeton, NJ (1943). Brunauer, S., and P.H. Emmett, J. Am. Chem. Soc, 62, 1732 (1940). Brunauer, S., P.H. Emmett, and E. Teller, J. Am. Chem. Soc, 60, 309 (1938). Clark, A., "The Theory of Adsorption and Catalysis," Academic Press, New York (1970). Emmett, P.H., and S. Brunauer, J. Am. Chem. Soc, 59,1553 (1937). Garcia Fierro, J.L.,S. Mendioroz, J. A. Pajares, and S.W. Weller, J. Catal., 65, 263 (1980). Herz, R.K., J.B. Kiela, and S.P. Marin, J. Catal., 73, 66 (1982). Hill, F.N., and P.W. Selwood, J. Am. Chem. Soc, 71, 2522 (1949) . Liu, H.C., and S.W. Weller, J. Catal., 66, 65 (1980). Liu, H.C., L. Yuan, and S.W. Weller, J. Catal., 61, 282 (1980). Maggiore, R., N. Giordano, C. Crisafulli, F. Cast i 11 i, L. Solarino, and J.C.J. Bart, J. Catal., 60, 193 (1979). Massoth, F.E., J. Catal., 30, 204 (1973). 114

PAGE 122

115 Miller, D.J., Ph.D. Dissertation, University of Florida (1982). Miller, D.J., and H.H. Lee, AIChE J., 30, 84 (1984). Millman. W.S., M. Crespin, A.C. Crillo, Jr., S. Abdo, and W.K. Hall, J. Catal., 60, 404 (1979). Parekh, B.S., and S.W. Weller, J. Catal., 47, 100 (1977). Parekh, B.S., and S.W. Weller, J. Catal., 55, 58 (1973). Phipps, A.M., and D.N. Hume, J. Chem. Educ, 45, 664 (1968). Ricca, F., "Adsorp ti on-Desorp ti on Phenomena," Academic Press, New York ( 1972) . Ross, S., and J. P. Oliver, "On Physical Adsorption," J.W. Wiley, New York ( 1964) . Sarakany, J., and Richard D. Gonzalez, J. Catal., 76, 75 (1982). Segawa, K.I., and W.K. Hall, J. Catal., 63, 447 (1983). Sinfelt, J., Chem. Eng. Prog., 63, 16 (1967). Spenedal, L., and M. Boudart, J. Phys. Chem., 64, 204 (1960). Srinivasan, R., H.C. Liu, and S.W. Weller, J. Catal., 57, 87 (1979). Stencel, J.M., L.E. Makovsky, and T.A. Sarkus, Submitted to J . Catal ., ( 1983) . Vaylon, J., and W.K. Hall, J. Catal., 84, 216 (1983). Wagner, CD., W.M. Riggs, L.E. Davis, J.F. Moulder, and G.E. Muilenberg, "Handbook of X-Ray Photoelectron Spectroscopy," Perk in Elmer Corporation, Minnessota (1979). Wanke, S.E., B.K. Lotochinski, and H.C. Si dwell. Can. J. Chem. Eng., 59. 357 (1981). Weller, W.S., and Sterling E. Voltz, J. Am. Chem. Soc, 76, 4695 (1954). Young, D.M., and A.D. Crowell, "Physical Adsorption of Gases," Butterworth, London (1962).

PAGE 123

BIOGRAPHICAL SKETCH The author was born and raised in Lahore, Pakistan. He earned his B.Sc. degree in chemical engineering from the Unversity of Engineering and Technology, Lahore, Pakistan, in 1979. He came to the United States of America in September 1979 to join the Tennessee Technological University, Cookeville, Tennessee, where he took his masters degree in chemical engineering, graduating in August 1981. In August 1981, he came to Gainesville and has been working on a Ph.D. degree in chemical engineering at the University of Florida. He is interested in conducting applied research in the field of reaction engineering and catalysis. 116

PAGE 124

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and in quality, as a dissertation for the degree of Doctor of Philosophy. Hong H. Leer Chairman Associate Professor of Chemica Engi neeri ng I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and in quality, as a dissertation for the degree of Doctor of Philosophy. r BTHofTuWd Ga Associate Professor of Chemical Engi neeri ng I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and in quality, as a dissertation for the degree of Doctor of Philosophy. Geral d B~! Westermann-Cl ark Assistant Professor of Chemical Engi neeri ng

PAGE 125

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and in quality, as a dissertation for the degree of Doctor of Philosophy. Jqhn P. O'Connell Professor of Chemical Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and in quality, as a dissertation for the degree of Doctor of Philosophy. Er i c R . Allen Professor of Environmental Engineering Sciences This dissertation was submitted to the Graduate Faculty of the College of Engineering and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. May, 1985 iLljq . ^. .'UMo Dean, College of Engineering Dean, Graduate School

PAGE 126

>.