The multiple equilibrium analysis model and its application to the study of adsorption


Material Information

The multiple equilibrium analysis model and its application to the study of adsorption
Physical Description:
xi, 189 leaves : ill. ; 29 cm.
McGilvray, John Michael, 1968-
Publication Date:


Subjects / Keywords:
Adsorption   ( lcsh )
Porous materials   ( lcsh )
Silica   ( lcsh )
Chemistry thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Chemistry -- UF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1997.
Includes bibliographical references (leaves 183-188).
Statement of Responsibility:
by John Michael McGilvray.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 028045230
oclc - 37824078
System ID:

This item is only available as the following downloads:

Full Text






Copyright 1997


John Michael McGilvray


I would like to thank the many people who have made my graduate career at the

University of Florida so memorable. I would especially like to recognize Professor Russell

S. Drago for his encouragement and guidance over the past few years, and Mrs. Drago for

her warm hospitality at the many Drago group functions.

There are many group members, past and present, I wish to thank for contributing

to my learning experience while at the University of Florida. Among these members I

wish to thank are Dr. Douglas S. Burs, for his patience, guidance, support, and above all

his friendship, and to his lovely wife Imogene for her wonderful dinners, Dr. David Singh

and Dr. Phil Kaufman for helping me start my research, Dr. Michael Robbins for those

wonderful afternoons riding horses in Newberry, FL, Scott Kassel and Ed Webster for

their help and support, and Andrew Cottone for being a trusted friend and fishing partner.

I would like to give special thanks to Dr. Samuel Colgate and Mr. Joe Shalosky

for their help in the machine shop when I needed to repair or build a piece of equipment. I

have grown to admire the quality and craftsmanship that these two gentlemen put into

their work, and I can only wish to achieve half of the expertise they possess.

I would also like to acknowledge Dr. Norris Hoffman and Mrs. M.R. Andrews at

the University of South Alabama and the other people who helped me obtain admittance

to the University of Florida.

Most importantly I would like to acknowledge the support of my wife, Victoria,

who always made the bad days a little better. Her kind thoughts and deeds were always

welcome and appreciated after a rough day at the lab. To my mother- and father-in-law

for all their help and support and for giving me such a wonderful wife. I would also like

to thank my father-in-law for his boat which always provided me an escape from the


Above all, I would like to acknowledge the support of my parents who always

taught me to strive for what I want and to never give up. To my parents I owe everything

and to them I dedicate this dissertation as a small token of my love and appreciation for all

they have done for me.


ACKNOWLEDGMENTS ......................................................................................iii

LIST OF TABLES ......................................................................................... vii

L IST O F FIG U R E S ...................................................................... ......................... viii

A B ST R A C T ............................................................................ ................................. x

CHAPTER 1: INTRODUCTION ................................................... ..........................

MULTIPLE EQUILIBRIA ANALYSIS MODEL ..........................................10

Introduction ....................................................................... ................... 10
Experim ental ....................................................... .......... ................... 14
R results and D discussion .............................................. .................... ....16
C conclusion .......................................... .................................................. 57


Introduction ................... ...... ...................................................60
Experim ental.................................................................................... ...................62
Results and Discussion ........... ................................................64
C conclusion .................................................. .......................................... 77

DERIVED SOLID ACIDS .......................................... .................... .... 79

Introduction ................................ ................................................... 79
E xperim ental ..................................................................... ....................85
R results and D discussion .............................................. .................... ....90
C conclusion ....................................................................... ................... 107

AD SORPTION DATA .................................................... ...................109

APPENDIX B: ADSORPTION DATA .......................................111


APPENDIX D: LIQUID ADSORPTION DATA ............................................177

LIST OF REFERENCES ..................................................... 183

BIOGRAPHICAL SKETCH ..................................... .... ......................189


Table MB

2-1 Physical Properties of Selected Gases........................................................ ........... 17

2-2 Physical Properties of Adsorbents................... ...... ..................18

2-3 Equilibrium Parameters for A-572 and PPAN ....................... ...................21

2-4 MEA Surface Areas and Pore Volumes for A-572 and PPAN ............................31

2-5 Equilibrium Parameters for HZSM -5....... ............. ..............................................34

2-6 Enthalpies and Entropies for A-572, HZSM-5 and Fisher Silica Gel ....................39

2-7 Equilibrium Parameters for Fisher Silica Gel ....................... .....................49

2-8 ZINDO Calculated Molecular Surface Areas and Volumes ................................53

2-9 MEA Surface Areas and Volumes for HZSM-5 and Fisher Silica Gel ..................55

3-1 Summary of Physical Properties of Selected Solutes and Solvents .......................65

3-2 Equilibrium Parameters for Adsorption from Cyclohexane by Silica Gel ..............68

3-3 Equilibrium Parameters for Adsorption from Cyclohexane by A-572 ..................73

4-1 N2 Porosimetry Data for Commercial and Sol-Gel Derived Silicas .....................91

4-2 N2 Porosimetry of Sulfated Silica Samples....................................................... 102

4-3 Cal-Ad Results for Sol-Gel Sulfated Silica ....................................... ........... 103

4-4 Preliminary Results for the Synthesis of MTBE by Sulfated Silica ...................106


Fire pae

1-1 Van der Waals Attraction as a Function of Pore Diameter.....................................4

1-2 Adsorption Potential in Micropores............................................5

2-1 Determination of Number of Multiple Adsorption Processes .............................13

2-2 Effects of Temperature on Adsorption ........................................ ............20

2-3 Process Resolved Isotherms for CI- Adsorption by A-572 ...............................24

2-4 In Ki vs. I/T(K) Plot for Adsorbates Studied with A-572 .................................... 27

2-5 -AH1 vs. Square Root of van der Waals Parameter [a] ........................................28

2-6 Adsorption of Non-Condensible Adsorbates by HZSM-5 ....................................36

2-7 Adsorption of Condensible Adsorbates by HZSM-5 ........................................37

2-8 In K1 vs. 1/T(K) Plot for Adsorbates Studied with HZSM-5 ................................40

2-9 Enthalpy and Entropy Correlation for HZSM-5.........................................43

2-10 -AHi vs. Square Root of van der Waals [a] Parameter.....................................46

2-11 Adsorption Isotherms for Non-Condensible Gases Studied with Silica Gel ........48

2-12 Process Resolved Isotherms for CH4 Adsorption by HZSM-5..........................52

3-1 M odel for Liquid Adsorption ................................... .................... ... 61

3-2 Adsorption Isotherms for Liquid Adsorptives Studied with Fisher Silica Gel .......67

3-3 In K1 vs. vdW[a]'" for Probes Studied with Silica ..........................................70

3-4 In K 1 vs Ee for Silica G el ......................................... ......... .... .............. 72


3-5 Adsorption Isotherms for Liquid Adsorptives Studied with A-572 .....................74

4-1 A cidic Sites of Silica G el ........................................ ................. ....................80

4-2 Bronsted Acidity of AICI3/ SG Catalyst ................................... ... ................ 81

4-3 Organic Bridged Silica Gels ................................................ ....................84

4-4 Pore Size Distribution of Davison Silica Gel ..................................................94

4-5 Pore Size Distribution of Fisher Silica Gel ..................................................... 95

4-6 Calorimetric Titration of Commercial and Sol-Gel Silicas ..................................96

4-7 Effect of Peroxide Wash on Davison Silica Gel ..............................................98

4-8 Effect of Peroxide Washing on Commercial Fisher Silica Gel ............................99

4-9 Effect of Peroxide Washing Sol-Gel Silicas .................................................100

4-10 Calorimetric Titration of Commercial and Sol-Gel Sulfated Silicas ..................101

4-11 TGA of Sulfated Silica ............ ....................................................104

4-12 Reaction oft-Butanol and Methanol to form MTBE .....................................105

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



John Michael McGilvray

August, 1997

Chairman: Dr. Russell S. Drago
Major Department: Chemistry

Porous solids are used extensively in many adsorbent and separation applications,

as well as serving as catalyst supports for many heterogeneously catalyzed reactions. Of

the many solids available, most specialty adsorbents are carbon or silica based. Many of
the carbonaceous adsorbents are derived from coal and wood chars, as well as being
synthetically produced from macrorecticular resins. The siliceous based adsorbents

include silica gel and zeolites. Silica, like carbon, is amorphous and provides a distribution

of pore sizes capable of adsorbing probe molecules of varying size and shape. Zeolites are

crystalline solids, some of which occur naturally while others are synthetically produced.
Crystallinity in a solid defines the limitation and application of an adsorbent to effectively

adsorb certain molecules while excluding others. Understanding those properties of a
solid which favor the adsorption of probe molecules is vital in selecting an adsorbent for

the appropriate application.
Currently there are several adsorption models in use, and many under

development, which characterize the physical properties of solid adsorbents, namely

surface area and pore size. One of the main challenges facing current adsorption models is

the ability to predict adsorption isotherms for new adsorbates based on the known

properties of the adsorbate and the adsorbent. Another challenge is the ability of these

adsorption models to be applicable to all adsorptives and all adsorbents.
Two commercial carbonaceous adsorbents and two silica based adsorbents are

examined for their ability to favorably adsorb selected probe molecules. Examination of
the adsorption data by the Multiple Equilibria Analysis (MEA) model has provided great

detail about the adsorption process. In an MEA interpretation, multiple adsorption

processes are found to contribute to the total adsorption isotherm. These multiple
adsorption processes exhibit different adsorption affinities which correspond to the

adsorption of probe molecules into pores of different dimensions. In conjunction with

equilibrium affinity, capacities for adsorption of a probe molecule can be transformed into

accessible surface areas and corresponding pore volumes. Enthalpies of adsorption for the
individual processes are calculated from the temperature dependent equilibrium constants.

Correlation of the enthalpies and equilibrium constants for the selected probe molecules

with the van der Waals [a] parameter is linear for those adsorbates undergoing

nonspecific, dispersion interactions. This affords the opportunity of the MEA model to

predict enthalpies and equilibrium constants, and in turn the adsorption isotherm, for

probes not yet examined.

Extension of the MEA model to liquid-solid adsorption equilibria is described for

selected adsorbates with a carbonaceous adsorbent and silica gel. Multiple equilibrium
constants are found indicating the presence of more than one adsorption process. For the
carbonaceous adsorbent, nonspecific interactions dominate, whereas silica undergoes

specific donor-acceptor interactions.

Preliminary results for the synthesis of a silica based acid catalyst and its potential

application for the production of methyl tertiary butyl ether (MTBE) are described.


From the early years of our civilization, porous materials have been used for

special applications. One of the earliest reported uses dates back approximately 1000

years ago when the Egyptians used carbon to purify drinking water for medicinal

purposes.' In 1756 Baron A.F. Cronstedt first discovered zeolites, aluminosilicates.2 He

noticed that, when these solids were heated, gases were evolved, and when cooled, they

adsorbed approximately the same volume previously lost. In 1777 Fontana noted that

freshly calcinated charcoal was capable of adsorbing many times its own volume of

various gases.' Since these early discoveries, porous materials have been utilized for a

wide range of applications including separation and catalytic processes. Understanding

the properties that most influence adsorption of an adsorptive by an adsorbent allows for

the appropriate selection of materials for adsorbent and catalytic applications.

Of the many adsorbents available, the most commonly used are carbon and silica

based. Porous carbonaceous adsorbents are derived from coal, peat, and wood chars, as

well as synthetically produced carbons made from macrorecticular resins.1'4 They are used

extensively as adsorbents for the removal of organic contaminants from waste water,s in

industrial flue gas scrubbers to minimize SO. and NO, pollutants,6 for military decon and

demil applications,7'8 as gas purifiers,9"as a means of storing fuels, fuel cells, and as

catalyst supports.

Metal oxides such as silica gel, SiO2, alumina, A1203, and zeolites are also used

extensively as adsorbents and catalyst supports.12-14 These solids offer a functionalized

surface capable of ion exchange for heavy metal removal from streams and aquifers.

Zeolites are recognized as strong solid acids and are used in the petrochemical industry as

alkylation, isomerization, and cracking catalysts.13",15"

The solids just mentioned, as well as others, have one thing in common; regardless

of their chemical composition, they are porous and they adsorb. However, not all solids

adsorb the same way, nor do they adsorb the same things. Silica gel and carbonaceous

adsorbents generally have no long range order to their structure and are considered

amorphous. Zeolites, unique in their composition, are recognized as being microporous,

crystalline solids. Crystallinity in a solid adds a limitation to the size of the adsorptive

capable of being adsorbed. This allows zeolites to function as size exclusion adsorbents

and catalysts." This feature is not common to amorphous materials where an adsorbate

molecule is capable of being adsorbed by a range of pore dimensions. The degree to

which a solid adsorbs a particular adsorptive is dependent on the porosity of the solid.

Porosity is a term used to describe a solid's pore size distribution and specific

surface area. IUPAC defines pores which are less than 20 A in diameter as micropores,

pores between 20 and 500 A as mesopores, and pores larger than 500 A in diameter as

macropores.13,17 Surface area, which is the general physical property used to describe

solids in the literature, is a function of pore size. Large quantities of smaller pores afford a

large internal surface area, whereas large pores result in a lower surface area.

The process of adsorption can be divided into two areas: chemisorption and

physisorption.'8 Chemisorption, in general, is the formation of a chemical bond between

the adsorbate and the adsorbent. The enthalpy of adsorption for a chemisorption process

is generally greater than 15 kcal mole-'. Physisorption, or reversible chemisorption, occurs

as a result of the attractive forces between the molecules of the solid and the gaseous or

liquid adsorbate. These forces are generally London dispersion forces and dipole-dipole

attraction potentials. Dispersion forces are a result of oscillations in the electron cloud of

the adsorptive and the adsorbent such that an attraction between the two occurs resulting

in adsorption.19'20 These interactions are weak and occur over short distances, decreasing

rapidly with distance (r6), Figure 1-1. The strength of the physisorption process will

depend on several factors, the first being the size and type of pores which are being

accessed. For those solids consisting of a range of pore sizes, the narrower the pore

diameter the greater the surface contact between the adsorbate and the adsorbent and the

greater the enthalpy of interaction,21 generally 5 to 15 kcal mole'. This was demonstrated

by Everett and Powl who calculated adsorption potentials for micropores in carbonaceous

adsorbents, assuming slit shaped and cylindrical pore geometry, and compared this to

adsorption on a graphitic plane surface." Cylidrical pores were found to experience a

greater adsorption potential than slit-shaped pores as a result of greater surface contact.

The maximum adsorption potential was found to occur in pores whose diameter closely

matched that of the adsorptive, Figure 1-2. As the pore diameter increased, the

adsorption potential approached that of a single graphitic surface. This indicates that the

energy of adsorption is closely related to the diameter of the adsorbing pore in relation to

Distance from Center of Molecule
to Surface of Interaction Measured
in Molecular Diameters

Figure 1-1: van der Waals Attraction at Molecular Dimensions. The attraction
between molecules decreases rapidly as the distance between molecules increases.


d = 1.90-a


d =1.33-a

-I 0 1


d =1.01 a

-1 0 1


Figure 1-2: Adsorption Potential in Micropores. The narrower the pore diameter
the stronger the adsorption potential. Here d is the diameter of the pore and a is the
size of the adsorptive.

the size of the adsorptive. As the pore diameter increases the adsorbate molecule

encounters a weaker surface attraction such that the enthalpy of interaction approaches

the enthalpy of vaporization of the adsorbate.

As mentioned previously, the surface area of a solid is one physical property which

is generally reported in the literature for describing the porosity of a material. There have

been many advances in the development of theoretical models to aid in the calculation of

surface area, but none have reached the point of predicting an adsorption isotherm based

on the known properties of the adsorbates and of the adsorbents.3 One of the earliest

models used to describe adsorption, which laid the foundation for many others to follow,

was the Langmuir equation (Equation 1-1).'323 Here P is the equilibrium pressure of A in

atmospheres above the solid, SA is the total amount of A adsorbed per gram of solid, n is

the number of sites available for adsorption, and K is the equilibrium constant describing

the affinity of A for being adsorbed by S. Examination of a plot of P/[SA] vs. P for a

homogeneous solid is a straight line of slope lln and an intercept of 1/nK. From the

determination ofn one can use the cross sectional area of the adsorptive to determine the

accessible surface area.3 Likewise, use of the molar volume of the adsorptive with n yields

the accessible pore volume.3

[SA] = nP] Equation 1-1

The Langmuir equation describes the adsorption of a molecule based on the assumption

that only a monolayer coverage is possible, that there are no adsorbate-adsorbate

interactions, and that the heat of adsorption is independent of surface coverage, meaning

that the solid has a homogeneous surface.1'" Clearly, it can be inferred from the

information provided above, that many, if not all, solids are heterogeneous in nature and

are comprised of a distribution of pore sizes and surface defects which exhibit different

adsorption potentials.

The BET (Brunauer, Emmett, and Teller) equation (Equation 1-2) was an attempt

to improve on the assumptions of Langmuir by including multilayer adsorption.1,3'25

X 1 (C-1)X
-- = -1 +C-I)X Equation 1-2
n(1- X) Cn. Cn.


K ep Q- Q-,
C =- exp

Here X= P/P,, and n is the total number of moles of adsorptive adsorbed per gram of

solid. The quantities n. and C are derived from a linear plot of X/n(l-x) vs. X. The BET

C constant is a complex quantity with contributions from the equilibrium constants, Kij,

and Kj,,d, and the enthalpies, Qon and Qmui, of adsorption for mono- and multilayer

formation. The enthalpy of adsorption for the multilayer is generally associated with the

enthalpy of vaporization of the adsorptive. The BET equation is generally considered the

standard for describing the surface area of solids;'26 however, it too has been subjected to

criticism due to its limited effective pressure range of 0.05 to 0.3 atmospheres and its

inability to accurately define the surface areas for microporous materials where it may

predict too large a surface area.'"

Other adsorption models such as the DR model proposed by Dubinin and

Radushkevich'"27 and the Horvath-Kawazoe2' model have examined the concept of pore

filling in order to establish an understanding of the pore size distribution of solids.

Although useful, these models involve extensive mathematical manipulation and the

empirical assignment of constants which at times may have to be altered in order to

explain or fit the adsorption isotherm. In summarizing the quandary of those interested in

understanding adsorption equilibria, Choudhary et al. state that no one adsorption model

(Langmuir, DR, BET, etc.) was capable of fitting their data for three individual probe

gases to the corresponding experimental isotherms for any one of the three zeolites

studied.9 From this statement it is clear that for all of the adsorption models in use, no

one particular model has yet to be found applicable for all solids at all pressures for all


Chapter 2 extends a newly reported adsorption model in an attempt to shed new

light on the theory of gas-solid equilibria.3'31 This model is capable of fitting the

adsorption data of an experimental isotherm with considerable accuracy. Although

similar to the Langmuir equation, the Multiple Equilibria Analysis, MEA, model is

designed to treat the adsorption of gaseous molecules as individual adsorption processes

and not as site specific interactions as in the Langmuir analysis. The equilibrium constants

represent the interactions of adsorbate molecules in pores that are too close in size to be

distinguished. As the name suggests, it is believed that multiple adsorption processes

occur in heterogeneous solids and that this model, along with the detailed information that

is obtainable, can lead to the prediction of adsorption isotherms for adsorbates not yet

tested on a particular solid. Chapter 3 is an extension of the MEA to liquid-solid equilibria

where the adsorption of several organic compounds from cyclohexane solutions by silica

gel and a carbonaceous adsorbent were studied.

Environmental and safety concerns for the handling and disposal of chemicals and

chemical waste has heightened the interest in porous materials serving as adsorbents and

catalysts. There are a number of benefits to designing solid acids for use in heterogeneous

catalysis. The first is the ease of product separation from the catalyst and the reactants.

The second is the ability of some solid acids to be regenerated, thus lowering the amount

of acid waste, as well as lowering financial expenditures. Chapter 4 presents the

development of a solid acid which is designed to reduce the amount of liquid acids, such

as H2S04 and HF, currently in use.



The adsorption, whether chemical or physical, of a gas, A(, by a solid, S.), can be

described by Equation 2-1.

S(,> + A(,) < SA(,) Equation 2-1

An equilibrium constant, K, can be written for this physisorption or chemisorption

interaction described in Equation 2-1 and rearranged to give Equation 2-2 .3,24

[SA]., Equation 2-2
1 + K[PI

Here, n is the capacity for adsorption, [SA], is the total amount of A adsorbed per gram of

solid, and P is the equilibrium pressure, in atmospheres, of A in equilibrium with the solid.

Though Equation 2-2 is similar in form to the Langmuir equation, the derivation with n

defined as a process capacity eliminates the assumptions of site specific interactions

associated with the Langmuir equation. The key feature of this equation is the treatment

of nonspecific adsorption by a distribution type of equilibrium constant that refers to a

process and not to a specific binding site. As a consequence, n refers to a process

capacity with the adsorbate undergoing interactions with the solid and neighboring

adsorbate molecules.

Porous materials contain a distribution of pore sizes and structural defects which

give heterogeneity to the solid.32 This heterogeneity gives rise to multiple equilibrium

expressions which represent the adsorption of a gas in the matrix of a solid. Each of these

equilibrium expressions can be described by Equation 2-2. Summation ofi expressions

like Equation 2-2, to accommodate i multiple adsorption processes, leads to Equation 2-3.

The capacities, equilibrium constants, and heats of adsorption define the processes that

contribute to the total adsorption isotherm.

[SA],= nK[P] Equation 2-3
i 1+K,[P]

Here [SA], is the total amount of A adsorbed by the solid, S, in the experiment in moles

per gram, ni, is the capacity for process i in moles, Ki is the equilibrium constant for

process i in atm"~ and P is the equilibrium pressure in atmospheres. From Equation 2-2, a

plot of [P]/[SA], vs. [P] for a one process adsorption isotherm would be linear;3'2

however, results at low temperatures may cause a curved isotherm, suggesting multiple

equilibrium processes.

Analysis of a data set with Equation 2-3 for a three process interpretation involves

solving for six unknowns, nl, n2, n3, Ki, K2, and K3. The ni,ad values for each process are

assumed to be temperature independent; therefore, each new data set, at a different

temperature, introduces only new Ki,,. values as unknowns. For a three process analysis,

the 6:1 ratio of unknowns to knowns at one temperature becomes 9:2 at two

temperatures, and 12:3 at three temperatures. Multiple temperature analyses decrease the

ratio of unknown to known parameters which leads to a better definition of the minimum

in the fit of the combined data set. In due process, the enthalpies for each adsorption

process, -AH~,a are calculated from the temperature dependent equilibrium constants

(Ki,~&) by plotting In Kla vs. I/T(K) (van't Hoff plot). Linearity of the graph affords a

check on the consistency of the fit parameters for the model employed.

To analyze data, a least squares minimization is used. In order to avoid locking

into false minima an estimate of the parameters is required. An isotherm of P/[SA] vs. [P]

is segmented into linear regions and a linear regression analysis is performed on the points

in each region to obtain preliminary n and K values. The slope of a linear portion is equal

to 1/n and the intercept is equal to 1/nK. The number of processes that will be initially

employed in the data fit is determined from the minimum number of straight line regions

one is able to distinguish in the total isotherm, Figure 2-1. The lowest pressure region is

associated with process 1, the next with process 2, continuing until the minimum number

of straight line regions is used to fit the entire isotherm. The preliminary n and K values

obtained for each straight line region are used as starting values in a simplex fitting

program designed to fit each of the multiple temperature adsorption data sets for a

particular probe simultaneously.

Only one set of n values is needed for the fit as the n's are assumed to be

temperature independent. Evidence of this temperature independence is apparent in many

of the isotherms where the calculated seed values are relatively the same for each



0.2 0.4 0.6 0.8 1.0

Pressure [atm]

Figure 2-1: Determination of the number of adsorption processes. The lines
through the data represent those points selected for a particular process.

corresponding n value at the different temperatures. For a three process fit, the initial

value of n is taken from the temperature set closest to the critical temperature of the

adsorptive. The value of n2 is taken from either the same temperature set or the next

highest, and n1 from the highest temperature set. It is at these temperatures that each

quantity is the best defined. The initial parameter set then contains one set ofn values and

a set of K values for each temperature. For the initial fit, each parameter is given a 10%/

bound and all of the temperature sets are fit simultaneously. After successive fits the

bound is reduced to 1% of the parameter. The bound limits the size of the step taken in

the search, but not the variation in the final parameter; therefore, the parameters are not

sensitive to the bounds selected. The parameters obtained from the initial fit are then

entered as the initial parameters for the next fit. The parameters from the second fit are

then entered as initial parameters into the third and each successive fit is done accordingly.

This procedure is used since several local minima are found during the first few fits.

Therefore, the fitting process is not assumed to be complete until all of the parameters

cease to vary from fit to fit. To assure oneself that the final minimum reached is not a

false minimum, the parameters are changed by several orders of magnitude and the fit is

repeated. If the final parameters reach the same values as before, one can feel confident in

the data analysis.


Adsorption analyses are conducted on a Micromeritics" ASAP 2000 gas analyzer

employing chemisorption software. A pressure table consisting of a minimum of 40

pressure points has been selected, Appendix A. The pressure table spans the range from

0.05 to 760 torr, with approximately 50% of the selected pressure points below 350 torr.

Pressure tolerances are set for 1% (10 torr pressure transducer) and ImmHg (100 torr

pressure transducer) with a 10 second equilibration time interval between selected

pressure points. The data are collected as unsmoothed data and exported to a spreadsheet

where the data in pressure (torr) and volume adsorbed (mL g1) are converted to pressure

in atmospheres and moles adsorbed.

Low temperature adsorption experiments (< 250C) were performed with the aid of

liquid N2 / solvent baths33 in Dewar flasks supported by the instrument's elevator unit.

High temperature adsorption experiments (> 25C) were performed using the heating

mantles supplied with the instrument's degas manifold with temperatures maintained to

within 1C. Adsorption measurements are typically performed at temperatures above

the critical temperature,T., of the adsorptive in order to avoid possible condensation of the

adsorptive in the pores of the solid.

The gases He and N2 (99.99% purity) were obtained from Liquid Air, Inc. The

gases CH4, CO, C2HI, C3H8, and dimethyl ether (DME) (99.99%) were purchased from

Matheson Gas Company. The gases SO2 and SF6 were purchased from Aldrich Chemical

in lecture bottle form. Properties of the selected gases are summarized in Table 2-1.

Sample preparation consisted of drying the adsorbent on the instrument's degas manifold

at elevated temperature and reduced pressure (<103 torr) for a minimum of 8 hours.

Typical sample weights were on the order of 0.25 to 0.3 g.

Carbon, hydrogen, and nitrogen analyses (C,H,N) were performed by the

University of Florida elemental analysis laboratory. The results are listed in Table 2-2.

For comparative purposes, surface area and pore volume data were obtained from the N2

isotherm at 77K using a Micromeretics' ASAP 2000 instrument. Surface areas were

determined using a five point BET calculation.'1,325 Micropore volume was determined

using the Harkins-Jura t-plot model with thickness parameters from 5.5-9.0 A.34 The BJH

desorption curve was used for calculating meso- and macropore volumes.35 Table 2-2 lists

the parameters found from N2 porosimetry for the various adsorbents.

Results and Discussion

Chemically different porous carbonaceous and silca based metal-oxide adsorbents

were studied with a series of probe gases. Ambersorb572 (A-572, lot #2201) is a

predominantly carbon containing pyrolized polymeric adsorbent, and PPAN (PLR 0246), a

pyrolyzed poly-acrylonitrile adsorbent. Both solids were supplied by the Rohm and Haas

Company. The metal oxide adsorbents consisted of a commercial silica gel (Fisher SG,

lot# 934403) and a commercial sample of the zeolite HZSM-5 (Si/Al = 50, PQ

Corporation). The gases used were selected to encompass a range of polarity,

polarizability, and donor-acceptor properties. Helium was used as a reference zero point

to determine the dead volume of the solids. The physical properties of the selected

adsorptives are listed in Table 2-1.

Results from the fitting of the multiple temperature data indicate that for a solid

where three processes are believed to be occurring, the use of temperatures above the

critical temperature provide useful information for the first two processes. At higher

Table 2-1: Summary of Gases and Physical Properties"

Probe MW Polariza Dipole Molar Critical Normal AH, van der Cross Kinetic
Gas g/mole ability" Moment Volume' Tempera Boiling kcal/mol Waals Sectional Diameterd
A3 Debyeb mLmol ture Point Const. Area A
"C C [a] [A2]
N2 28.01 1.74 0 35.4 146.9 -195.8 1.33 1.390 16.2 3.64
CO 28.01 1.95 0.13 34.9 140.2 -191.5 1.44 1.485 15.0 3.76
CH4 16.04 2.59 0 37.8 -82.60 -161.5 1.95 2.253 17.8 3.8
C2H6 30.07 4.47 0 52.6 32.3 -88.65 3.52 5.489 23.0 -4.3
C3Hs 44.10 6.29 0.084 75.0 96.7 -42.1 4.49 8.664 36.1 4.3
DME 46.07 5.16 1.3 127 -25 36.78 8.073
SF6 146.05 6.54 0 75.5 45.55 -63.8,b 4.08 26.7 5.5
SO2 64.06 4.28 1.63 43.8 157.5 -10.0 5.96 6.714 27.1 3.6

Table 2-2: Physical Properties of Adsorbents

Fisher SG HZSM-5 A572 PPAN
BET Surface Area 508 402 1222.7 873.7
m2g '
Pore Volume [mL g ']
Microporeb 0.02 0.15 0.45 0.33
Mesopore' 0.33 0.17 0.35 0.19
Macropore ---- ----- 0.20 0.10
CHN Analysis
%C 0 0.61 91.3 70.18
%OH 0.78 0.43 0.33 1.66
%N 0 0.25 0 5.31

a. Surface area and pore volume data are obtained from N2 porosimetry at 77K.
b. Micropore volumes calculated from the Harkins-Jura t-plot model.
c. Meso- and Macropore volumes calculated from Barrett-Joyner-Halenda (BJH) desorption data.

temperatures less adsorptive is taken up and the complete adsorption process spans a

larger range of pressures before they are complete. Process 3 is incomplete under

conditions where Process 1 and Process 2 are well defined. Since more adsorptive is

taken up as the temperature decreases, Figure 2-2, Process 1 determination will have a

large error as it may reach capacity in the first two or three pressure points. However, at

this temperature more of the data points have significant contributions from Process 3.

Therefore, Process 3 is best defined by low temperature data near the critical temperature

of the adsorbate. Process 1 is best defined at higher temperatures. Resolution of Process

2 appears to be good at all temperatures. Determination of the process parameters is not

considered to be adequate until the final process capacity is at least 70 % of the total

amount adsorbed at the lowest temperature.

Carbonaceous Adsorbents: A-572 and PPAN

The amorphous, carbonaceous adsorbents A-572 and PPAN were examined for

their abiltiy to adsorb various probe molecules. Table 2-3 summarizes the results of the

data fits for the adsorbents where n, is the capacity of the solid for the i11 adsorption

process in millimoles per gram of solid and K,a.,, the adsorption equilibrium constant in

atm"1, is a measure of the affinity of the probe to be adsorbed by the solid. The optimized

n and K values reported in Table 2-3 can be used in Equation 2-3 to solve for [SA], for

each process. Figure 2-3 shows the process resolved isotherms for CH4 adsorption by A-

572. The resolved isotherms illustrate the capacities for the individual processes and the

pressures at which maximum capacity for a process is reached. The initial slope of the

0.006- I j i I I '

0.005 0 750C
7 550C
S0.004- 400C
O 250C 9 E E
" 0.003 9 v 7

v [] [] [1

0.001 -E

0 0.2 0.4 0.6 0.8 1.0

Pressure [atm]

Figure 2-2: Effect of Temperature on the Amount of Adsorbate Adsorbed.

Table 2-3: Equilibrium Constants, n-values, Enthalpies, and Free Energies

N2 / A572
ni = 0.4630.006 n2 = 1.790.04 n3 = 5.30.2
-AHI = 4.800.01 -AH2 = 4.30.2 -AH3 = 3.00.2
TC K1 -AGI K2 -AG2 Ks -AG3
-93 88.150.02 1.6022 8.6260.005 0.7708 0.7520.001 -0.102
-43 4.50.1 0.69 0.720.03 -0.15 0.1110.009 -1.004
0 0.9220.002 -0.0441 0.13790.001 -1.0747 0.0460.0002 -1.674

nj = 0.50+0.01 n2 = 1.590.09 n3 = 3.60.2
-AHi = 4.80.1 -AH2 = 4.40.3 -AH3 = 2.700.1
TC Ki -AGI K2 -AG2 K3 -AG3
-93 84.650.02 1.5877 8.7230.007 0.7748 0.8130.003 -0.0739
-43 4.210.08 0.657 0.660.03 -0.19 0.120.01 -0.98
25 0.460.01 -0.46 0.0430.004 -1.86 0.0420.002 -1.87

CO / A572
n = 0.520.01 n2= 1.900.09 n3= 4.90.9
-AHI = 5.3+0.3 -AH2 = 4.30.3 -AH3 = 3.40.2
TC K1 -AG1 K2 -AG2 K3 -AG3
-93 2492 1.9740 19.80.5 1.0690 1.40.2 0.110
-77 602 1.60 5.80.4 0.69 0.50.2 -0.200
-62 26.40.6 1.372 3.10.2 0.4800 0.340.06 -0.460
-43 9.70.61 1.03 1.370.03 0.1450 0.160.01 -0.820

ni = 0.520.01 n2 = 1.900.09 n3 = 4.90.9
-AHI = 5.30.3 -AH2 = 4.30.3 -AH3 = 3.40.2
TC KI -AGI K2 -AG2 K3 -AG3
-93 2431 1.965 18.70.5 1.048 1.30.2 0.10
-43 8.80.3 0.99 1.100.08 0.0452 0.150.04 -0.88
25 0.700.03 -0.21 0.050.01 -1.8 0.0500.005 -1.78
ni in mmoles, Ki in atm"', -AHi in kcal mole''

Table 2-3 continued

CH4 / A572
nl = 0.940.01 n2 = 2.510.1 n3 = 51
-AHI = 6.20.4 -AH2 = 5.10.2 -AH3 = 4.00.3
TC Ki -AGI K2 -AG2 K3 -AG3
-77 359.60.6 2.2922 20.10.2 1.168 1.80. 1 0.22
-62 1002 1.93 7.90.6 0.87 0.80.3 -0.07
-43 23.70. 2 1.447 2.410.07 0.402 0.300.03 -0.55
0 4.1010.01 0.766 0.5050.004 -0.370 0.1030.002 -1.235

nt = 0.290.01 n2 = 1.400.05 n3 = 3.60.2
-AHi = 6.8110.06 -AH2 = 5.590.01 -AH3 = 4.700.03
TIC K1 -AGI K2 -AG2 K3 -AG3
-43 91.20.1 2.063 9.9810.02 1.051 0.9330.007 -0.0316
0 8.60.1 1.17 1.460.02 0.204 0.1860.009 -0.912
25 2.90.1 0.64 0.600.03 -0.30 0.090.01 -1.5
40 1.930.08 0.409 0.390.02 -0.58 0.0610.007 -1.74

C3H8 / A572
n = 0.1980.003 n2= 1.010.01 n3 = 2.50.1
-AHI = 123 -AH2 = 9.40.3 -AH3 = 8.30.6
125 118.80.7 3.7787 13.40.1 2.050 1.2010.05 0.145
150 29.50.4 2.844 6.800.07 1.611 0.680.03 -0.32
175 23.310.2 2.802 3.520.05 1.120 0.370.02 -0.90

n = 0.2210.004 n2 = 0.870.02 n3 = 1.510.1
-AH, = 12.60.3 -AH2 = 91 -AHL = 82
TC KI -AG, K2 -AG2 K3 -AG3
125 86.10.7 3.524 9.40.2 1.77 1.20.1 0.15
150 33.010.2 2.940 5.260.06 1.396 0.740.04 -0.25
175 15.00.3 2.411 2.500.07 0.815 0.340.04 -0.97
ni in mmoles, K, in atm'', -AHi in kcal mole'

Table 2-3 continued

C2H6 / A572
n, = 0.3180.005 n2 = 1.42+0.03 n3 = 3.70.3
-AHI = 10.20.4 -AH2 = 8.10.2 -AH3 = 6.810.2
Tr KI -AGi K2 -AG2 K3 -AG3
40 176.70.6 3.2186 15.00.1 1.683 1.350.05 0.185
55 741 2.81 7.60.2 1.33 0.750.08 -0.19
70 42.00.4 2.548 4.810.09 1.072 0.530.03 -0.43
100 13.90.3 1.949 1.880.07 0.466 0.230.03 -1.08

S02 / A572
ni = 0.932210.0005 n2 = 4.70.1 n3 = 62
-AHI = 9.40.2 -AH2 = 8.10.1 -AH3 = 8.00.1
Tr K, -AGI K2 -AG2 K3 -AG3
25 360560 3.5 11.80.3 1.463 2.80.2 0.61
40 14035 3.06 6.10.9 1.13 1.40.9 0.22
55 6615 2.7 3.40.8 0.80 0.80.6 -0.1
75 304 2.35 1.60.5 0.32 0.40.4 -0.6

nj = 0.6580.00002 n2= 0.690.002 n3= 4.010.2
-AHH = 7.12.4 -AH2 = 1.10.1 -AH3 =4.80.4
TIC K, -AGI K2 -AG2 K3 -AG3
25 2E6lE6 5.9 20960 3.2 14.20.3 1.6
40 2E34E4 4.8 82.620 2.7 6.70.4 1.2
55 9008600 4.4 36.25.5 2.3 3.40.1 0.8
75 300800 3.9 14.02.3 1.8 1.50.9 0.3
ni in mmoles, Ki in atm'l, -AHi in kcal mole"'

resolved processes provides some indication as to the affinity for adsorption in that


The equilibrium constants are temperature dependent and are a factor of the

inherent physical properties (polarizability, van der Waals, etc.) of the adsorbent and

adsorptive. For small pores, the equilibrium constant will be greater than that


0.004- Process 1 ..........
Process 2--- -
S0.003 Process 3-----

3 0.002 -
-f -' --"- --------- -2-

0.001 ----

0 < a s .. .,
0 0.2 0.4 0.6 0.8 1.0
Pressure [atm]

Figure 2-3: Process Resolved Isotherms for CH4 Adsorption by A-572.

observed for larger pores due to greater interaction of the adsorbate with the walls of the

solid (see Chapter 1, Figure 1-2). Larger pores allow more freedom within the pore

structure such that an adsorbate molecule moves from one wall to another. Figure 2-4

shows the In KI,,. vs. 1/T(K) plots for the adsorbates studied with A-572. The

creditability of the model can be appreciated in the linearity of these adsorbates and the

calculated enthalpies whose values are indicative of physisorption processes.18

An interesting trend is observed among the solids studied. As the van der Waals

[a] parameter of the adsorbates increases so do the equilibrium constants for adsorption.

The van der Waals attraction parameter [a] is a measure of the strength of interaction of

like molecules for one another. The square root of [a] should characterize the relative

attraction of gas molecules for the solid surface. A plot of In K1. or -AH,,.d. vs. vdW

[a]'R for A-572 is linear for the probes examined, Figure 2-5. This not only provides

support for the meaning of the fit parameters in the context of the model used, but also

permits one to obtain an estimate of the equilibrium constant or enthalpy for the

adsorption of molecules not yet studied. This can be of practical utility when one is

selecting adsorbents for specific adsorption applications. Deviations from linearity will

indicate contributions other than nonspecific interactions present in the adsorption

process. A least squares analysis of such plots yields a coefficient which is representative

of the overall surface attraction for a particular process. With the exception of SO2 on

PPAN, all of the adsorbates studied fall on or near the best fit lines given in Equations 2-4

through 2-8.


-AH, = 3.0 ( 0.3) a"2 + 1.54 ( 0.53) R2 = 0.96 Equation 2-4

-AH2 = 2.7 ( 0.1) a'R + 1.31 ( 0.29) R2 = 0.98 Equation 2-5

-AH3 = 3.24 (0.15) a"2 0.75 ( 0.31) R = 0.98 Equation 2-6


-AH, = 3.32 (0.33) ala + 1.16 ( 0.61) R2= 0.98 Equation 2-7

-AH2 = 2.96 ( 0.57) ala + 0.58 ( 1.10) R2 = 0.93 Equation 2-8

Within the calculated error of the coefficients, the values exhibit a relatively uniform

surface attraction for the individual processes for A-572 and PPAN, indicating that the

magnitude of the equilibrium constants is determined by the overall size of the pore in

relation to the adsorbate. Specific donor-acceptor interactions ofadsorbate molecules

with a solid surface are expected to give rise to deviations in the van der Waals plots if

these interactions are stronger than those in the gaseous adsorbate. Therefore, we

conclude that the solid-adsorbate interactions for the systems described in the above

equations represent nonspecific interactions for those pores which are on the order of

molecular dimensions for the adsorbates studied.

It is significant that the enthalpies for N2, CO, CH4, C2H6, and C3H8 fall on a

straight line for Process 1 even though our analysis of the capacities indicated that a

different distribution of micropores is utilized for the different gases. Thus, Process 1

utilizes pore widths of molecular dimensions. Process 2 involves dimensions up to two

0.002 0.003 0.004 0.005


Figure 2-4: In K vs. 1/T(K) Plot for the Gaseous Probes Adsorbed by A-572.


14 I I I I I I I i I
+ N2

12- 0 CO

10 C6 c -
v C3H8

4 -

1.0 1.5 2.0 2.5 3.0

van der Waals [a] 12

Figure 2-5: -AHt vs. Square Root of van der Waals [a] for A-572.

molecules at their van der Waals separation interacting with the walls and each other.

Process 3 involves larger pores in which molecules interact strongly with one wall and

weakly or not at all with the opposite wall. In this manner, a given process i is

fundamentally the same for all probes studied, but different dimension pores are utilized

for the ith process of each probe.3

The relative importance of dispersion and nonspecific dipole-dipole interaction can

be inferred from plots of In K., or -AHi. versus polarizability (a) of the adsorbate.

Comparisons of these plots for different adsorbates describe the relative effective surface

polarizability of the solid. Equations 2-9 through 2-13 represent the equations for the

best fit lines for N2, CO, CH4, C2H6, and C3Hg for A-572 and PPAN. Note that the

coefficients again are within experimental error for each of the processes listed indicating a

uniform surface polarization.


-AHI = 1.19 ( 0.7) a + 3.00 ( 0.6) R2 = 0.94 Equation 2-9

-AH2 = 1.07 ( 0.9) a + 2.58 ( 0.3) R2 = 0.98 Equation 2-10

-AH3 = 1.25 ( 0.07) a + 0.90 ( 0.3) R2 = 0.99 Equation 2-11


-AH = 1.30 (0.12) a + 2.71 (0.4) R2= 0.98 Equation 2-12

-AH2 = 1.15 ( 0.24) a + 2.02 (0.9) R2 = 0.92 Equation 2-13

The n values determined on most gas-solid equilibria studied to date indicate that

the three processes are occurring in the micropores of these solids. Information similar to

that presented in Table 2-3 can be used to estimate the accessible micropore surface area

of the solid based upon the cross sectional area38 or kinetic diameter13 of the adsorptive.

Pore volumes for the individual processes can be determined from the molar volume of the

adsorptive. The adsorptives studied vary in size such that the pores for Process 1 of one

adsorptive will not be the same for a larger adsorptive. Therefore, the process capacities

of the MEA model are relevant to the micropore distribution of porous materials that are

characteristic of the specific adsorptive. Table 2-4 lists the surface areas and volume

capacities for the processes of A-572 and PPAN. The MEA surface areas are

considerably lower than the reported BET surface area of-1200 m2 g". This is believed

to be a result of condensation of N2 at 77K in the larger pores of the solids. This also

would allow for the pore volumes calculated by MEA to be much smaller than that

determined from standard nitrogen porosimetry.

Silica Gel and Zeolites

Silica gel (amorphous) and HZSM-5 (crystalline) afford samples of similar

composition and vastly different porosity that can be compared to the amorphous carbons.

Silica gel consists mainly of mesopores, while N2 porosimetry suggests that HZSM-5

contains both micro- and mesopores, Table 2-2. X-ray data indicate that HZSM-5 is a

microporous material contradicting N2 porosimetry results. Nonpolar adsorbates of

varying dimensions and shapes were selected to investigate the potential of the MEA

model in discerning porosity differences.

Table 2-4: Process Resolved Pore Volumes and Surface Areas for A572 and PPAN

N2 N2
Process mmol ads ml ads area[mg '] mmol ads ml ads area[m g"]
1 0.463 0.016 45 0.499 0.018 49
2 1.790 0.063 175 1.588 0.056 155
3 5.335 0.189 520 3.585 0.127 350
Total 7.588 0.269 740 5.672 0.201 553

Process mmol ads ml ads area[m2g'] mmol ads ml ads area[m2g*]
1 0.520 0.018 47 0.570 0.020 51
2 1.898 0.066 171 1.802 0.063 163
3 4.932 0.172 445 3.642 0.127 329
Total 7.351 0.257 664 3.014 0.210 543

Process mmol ads ml ads area[m2''] mmol ads ml ads area[m2g]
1 0.938 0.035 100 0.285 0.011 31
2 2.492 0.094 267 1.401 0.053 150
3 5.394 0.204 578 3.628 0.137 389
Total 8.824 0.334 945 5.314 0.201 569

Process mmol ads ml ads area[m2g'] mmol ads ml ads area[m2g]
1 0.198 0.015 43 0.221 0.017 48
2 1.007 0.075 219 0.874 0.066 190
3 2.488 0.187 541 1.529 0.115 332
Total 3.693 0.277 803 2.624 0.197 570

Process mmol ads ml ads area[mjg-]
1 0.318 0.017 44
2 1.422 0.075 197
3 3.718 0.196 515
Total 5.459 0.287 756

Adsorption Isotherms

For HZSM-5, and related solids with the MFI structure,39 the straight (elliptical

shaped, 5.4 x 5.6 A) and zig-zag (almost circular shaped, 5.1 x 5.4 A) channels differ

slightly in dimension, so at least two processes are expected." The calculated ni,d. values

and the temperature dependent Kid. values for all systems studied are listed in Table 2-5.

The symbols in Figure 2-6 and Figure 2-7 show representative adsorption isotherms for

HZSM-5 (Si / Al = 50) at two of the several temperatures required for a MEA analysis.

The temperature dependence of Kl,a and K2,d. gives the enthalpies (-AH~d.) and

entropies (-ASi.4) for the processes via the van't Hoff equation (InKi = -AHi / RT + ASi /


N2 InK1 = 2468.89(1/T) 10.4877; InK2 = 1557.767(1/T) 7.99366

CO InK = 3072.531(1/T) 11.0588; InK2 2606.553(1/T) 11.0546

InK3 = 1553.595(1/T) 7.8888

CH4 InK1 = 2585.9(1/T) 9.19493; InK2 = 2286.986(1/T) 10.4992

C2H6 InK1 = 4275.89(1/T) 11.4402; In K2 = 4643.942(1/T) 17.5608

C31s InKi = 4856.225(1/T) 10.3288; In K2 = 6234.666(1/T) 15.7839

SF6 InKi = 3819.4(1/T) 8.30587; In K2 = 4803.92(1/T) 13.7946

DME InK= 11247.42(1/T)- 17.4082; InK2 = 6523.715(1/T) 11.8505;

In K3 = 5320(1/T) 12.2464

These results are summarized in Table 2-6 and the In K1,.~d vs. 1/T(K) plots for HZSM-5

are shown in Figure 2-8. The linearity of these plots, and enthalpies, with magnitudes

expected for physisorption processes, indicate that meaningful parameters result from the


In our examination of the carbonaceous adsorbents mentioned in the previous

section, the most exothermic process, Process 1 was assigned to adsorption in the smallest

channels/pores of the solid, and Process 2 to the larger channels/pores. Comparison of the

kinetic diameters of the probes (Table 2-1) with the average channel diameter of 5.5 A for

HZSM-5 indicates that the channel is -1.0 to 1.5 times as large as the kinetic diameter of

the gas molecules studied. Thus, in contrast to carbon where the adsorbate selects those

pores of comparable dimensions for Process 1 from the distribution of pore sizes available,

Process 1 in zeolites corresponds to different adsorbate-wall interactions for the different

size adsorbate. For HZSM-5, and similar zeolites, the channels of the solid approximate

cylinders. For small molecules, i.e. N2, adsorbate-adsorbent interactions will occur with

one side of the channel. Based on the kinetic diameter of N2, and the diameter of the

channel, two molecules cannot occupy opposite positions on adjacent walls. A single N2

molecule is expected to adsorb on the channel surface providing the maximum surface

interaction and will interact weakly with the rest of the cylinder wall. As the adsorbates

increase in size, interaction with the distant walls of the channel increases becoming a

maximum when the molecular diameter is that of the cylinder. Thus, the increasing values

of the enthalpies for Process 1 (Table 2-5) as the size of the adsorbate increases is due to

Table 2-5: Equilibrium Constants, n-values, Enthalpies, Free Energies

N2 / HZSM-5
n,= 1.50.01 n2 = 1.30.01
-AH, = 4.90.3 -AH2 = 3.10.4
Temp'C K, -AG, K2 -AG2
-93 22.70.8 1.116 2.30.3 0.292
-43 1.50.01 0.188 0.20.01 -0.704
0 0.30.01 -0.731 0.80.01 -1.344
25 0.090.04 -1.423 0.090.05 -1.445

n, = 0.60890.0001 n2 = 1.2220.005 n3 = 1.70.4
-AH, = 6.10.08 -AH2= 5.20.3 -AH3 = 3.10.5
Temp'C K, K2 K3
-93 4179 284 25
-43 8.70.6 1.50.3 0.20.3
-16 2.560.07 0.460.03 0.150.04
0 1.2340.004 0.1780.002 0.1510.002

CH4 / HZSM-5
nl = 0.20.01 n2 = 0.920.06
-AHl = 5.10.03 -AH2 = 4.50.1
Temp'C Ki -AGI K2 -AG2
-84 8617 1.672 57 0.605
-42 8.140.07 0.962 0.550.07 -0.276
-16 2.310.01 0.427 0.20.02 -0.846
0 1.280.01 0.133 0.130.03 -1.123

C3Hs / HZSM-5
n- = 0.81890.0007 n2 = 0.850.03
-AHl = 9.70.5 -AH2 = 12.40.9
TempoC K, -AGI K2 -AG2
75 384 2.515 7.912.1 1.429
125 5.980.9 1.414 0.90.8 -0.083
150 3.580.12 1.072 0.440.09 -0.7308
175 1.580.05 0.407 0.130.04 -1.816
ni in mmoles, Ki in atm'1, -AHj in kcal mole'

Table 2-5 continued

C2H6 / HZSM-5
nt =1.920.06 n2 =0.65
-AHI = 8.50.2 -AH2 = 9.21.7
TempC K1 -AG1 K2 -AG2
-16 18910 2.676 232 0.354
0 682 2.288 0.77 -0.193
25 17.60.2 1.698 0.20.6 -0.953
40 9.410.02 1.394 0.040.08 -2.002
50 5.280.02 1.068 0.020.09 -2.511
75 2.610.01 0.663 0.030.05 -2.425

SF6 / HZSM-5
n, = 0.560.005 1n2 =1.20.1
-AH, = 7.60.3 -AH2 = 9.50.006
TempC Ki -AGi K2 -AG2
50 34.110. 2 2.265 2.940.09 0.683
65 19.60.2 1.998 1.50.1 0.2723
80 12.570.06 1.777 0.830.03 -0.156

ni = 0.6160.05 n2 = 0.153 n3 = 155
-AH1 = 22.44.5 -AH2 = 130.7 -AH, = 10.61.3
TempC KI K2 K3
150 83246 3526 13
175 30563 1613 0.71
200 4925 7191 0.325
ni in mmoles, K, in atm'', -AH, in kcal mole"

the increased polarizabilities of the larger molecules, and also to increased interaction of

the adsorbate with the surrounding walls.

With the carbonaceous adsorbents, the term process has been used to describe

collectively the affinity (Ki,.ad) and the strength (-AHi,.d,) of adsorption of a gas molecule

into a particualr size pore distribution. Process 1 would correspond to pores that were on

the order of the size of the gas molecule which in turn had the greatest enthalpy of




0.0005 V CO

0 0.2 0.4 0.6 0.8 1.0

Pressure [atm]

Figure 2-6: Adsorption Isotherms for the Non-Condensible Gases Studied with HZSM-5.



m ~- f ^ v^ Ethane 750C

0.0005 m Propane 750C

v SF, 800C

0 0.2 0.4 0.6 0.8 1.0 1.2

Pressure [atm]

Figure 2-7: Adsorption Isotherms for the Condensible Gases Studied with HZSM-5.

interaction (-AHI), Process 2 for pores that were 1.5 times as large, and Process 3 for

pores that were at least twice the size of the gaseous adsorbate. As described above,

zeolites, or solids with a defined pore structure, present problems with this interpretation

of the processes. We shall, therefore, define nomenclature to describe the adsorption of a

molecule. The order of pore filling, which depends upon K, will be designated by the term

Process with Process 1 corresponding to the pore that fills with greater ease. The term

Type will be used to describe the matching of the adsorbate and pore dimensions. A Type

1 adsorbate is one in which the molecular dimensions of the adsorbate match the pore. A

Type 2 adsorbate interacts more strongly with one wall than the rest of the pore, while a

Type 3 adsorbate interacts strongly with one wall and weakly with the rest of the pore

surface or with a molecule on the opposite surface. In contrast to HZSM-5, the pores of

carbonaceous A-572 consist of parallel graphitic sheets or planes. The adsorbate-

adsorbent interactions occur along the surface of these planes, with adsorbate-adsorbate

interactions also occurring parallel to the planes. The consequences of these differences

on adsorption by HZSM-5 compared to carbon will be elucidated.

The difference in the process, type designation is manifested in the thermodynamic

data for adsorption for HZSM-5. Consider the two processes for the larger adsorbates

SF6, ethane and propane, which have approximately the same dimensions as the channels.

The process with the more -AH corresponds to the process with the smaller K at elevated

temperature. The calculations of Everett and Powl22 indicate that the strength of the

Table 2-6: Enthalpies and Entropies for A572, HZSM-5, and FSG

-AHI2.3 kcal/mol -AS/R1,2
A572 4.8 0.01, 4.3 0.2, 17.8 0.3, 19.6 0.8,
3.00.2 17.1 0.7
HZSM5 4.9 0.3, 3.1 0.4 21 0.5, 16 0.9
FSG 4.2 0.4, 2.9 0.1 20.01 1.6, 18.4 0.3
A572 5.3 0.3, 4.3 0.3, 18.5 1.3, 18.3 1.4,
3.4 0.2 18.7 0.8
HZSM5 6.1 0.08,5.2 0.3, 22: 0.2, 22 0.2, 150.2
FSG 4.1 0.2, 3.6 0.2 18.7 0.7, 21 0.72
A572 6.2 0.4, 5.1 0.2, 20.1 1.8, 20.2 0.9,
4.0 0.3 19.2 1.4
HZSM5 5.1 0.03, 4.5 0.1 18 0.3, 210.2
FSG 4.6 0.1, 3.2 0.1 19.5 0.3, 18.9 0.5
A572 10.2 0.4, 8.1 0.2, 20.8 1.3, 20.01 0.8,
6.8 0.2 20.9 0.9
HZSM5 8.5 0.2; 9.21.7 22.7 0.7; 34.95.7
FSG 5.0 0.4; 5.60.3 16.6 1.4; 22.40.8
A572 12 3, 9.4 0.3, 20.2 10.2, 18.6 1.1,
8.3 0.6 20.5 1.1
HZSM5 9.7 0.5, 12.4 0.9 20.5 1.3; 31.4 1.3
FSG 4.9 0.8, 6.8 0.8 12.02 2.02, 24 2
HZSM5 7.6 0.3, 9.5 0.006 16.5 0.8; 27.4 0.02
HZSM5 22.4 4.5, 13 0.7; 34.6 10.1; 23.6 1.5;
10.61.3 24.33.0

6 ON
4-. VCH4
4 OH4C2H6 /

M 2- oSF6

0 0

I I I i l i l a i i I i I l ,
0.001 0.002 0.003 0.004 0.005 0.006


Figure 2-8: In K, vs. 1/T(K) for the Gaseous Adsorbates Studied with HZSM-5.

adsorptive-solid interactions should be based on enthalpies and not upon the equilibrium

constants which contain entropic contributions." With the larger enthalpy assigned to the

smaller zig-zag channel, the entropy associated with constraining the larger molecule in

the small channel causes the equilibrium constant for adsorption for the smaller channels

to decrease. Consistent with these proposals, June et al. report that the channels of the

zeolite exhibit constraining effects upon lengthening hydrocarbon chains, such that these

molecules are forced to take on a more elongated shape.4143 In zeolites with small

channels of fixed dimensions, the entropic factors associated with the loss of rotational

freedom when larger adsorbates enter the channels leads to a smaller equilibrium constant

for the more strongly adsorbed molecule.

For the smaller adsorbates, N2 and CH4, the more negative enthalpies for Process

1, are also manifested in a larger equilibrium constant than those for Process 2.

Adsorption of CO was best defined by a three process interpretation of the adsorption

data. Process 1 is attributed to weak hydrogen bonding with the acid sites of HZSM-5.

Although CO has a weak dipole, correlation of ni for CO with nl for DME indicates that

both gases are interacting with the same number of hydrogen bonding sites. CO is weakly

basic compared to DME and the magnitude of the DME interaction is evident in the

enthalpy for the first process. Processes 2 and 3 for CO correspond to similar interactions

observed for Processes 1 and 2 of N2.

In order to gain further insights from the thermodynamic data, a plot of the

entropy vs. enthalpy of adsorption was constructed and is shown in Figure 2-9. When

data for all the processes are combined a poor correlation results. However, when the

data are divided into two sets corresponding to adsorption in the smaller and larger

channels, better correlations are found. Smaller enthalpies are expected for adsorption of

molecules in larger channels as generally observed. However, the need for two different

enthalpy/entropy plots indicates different entropies associated with filling the two

channels. This could result from the adsorbate leading to different losses of vibrational

freedom for the silicon and aluminum atoms comprising the walls of the different size


The following adsorptives, with the process listed in parentheses, are assigned to

the smaller channels: N2(1), CO(2), CH4(2), C3HI(2), SF6(2). The following are assigned

to the larger channels: N2(2), CO(3), CH1(l), SF6(I), and C3Hg(1). Least squares lines

are constructed for these two assignments in Figure 2-9. The smaller channel plot gives

an excellent correlation (R2 = 0.99). The Type of interaction is different for N2 and CO

than for propane and SF6. Thus the slope of the line contains an enthalpy and entropy

component from the dispersion interaction and from changing the Type of interaction.

Those adsorbate processes assigned to the larger channel span a range of only four

entropy units and give a poorer enthalpy entropy plot (R2 = 0.66)

For methane, the Process 1 point with the larger enthalpy falls on the large channel

plot while Process 2 falls on the small channel plot. Contrary to expectation, with either

assignment the enthalpy for methane binding is lower than that for N2. The low -AS

causes methane to pack in the small channel in a manner that gives rise to a lower enthalpy

of interaction. One possibility would involve interaction of the edge instead of the face of

- I I I I I I I I i I I I I

S *Small Channel HZSM-5
Large Channel HZSM-5
A Small Channel FSG

0 ~ 0
0 c

,I I I I I I I I I I

-AH [kcal/mol]

Figure 2-9: Entropy and Enthalpy Correlation for the Large and Small Channel
of HZSM-5. The Correlation for the Small Channel of Silica is also shown.

the tetrahedral methane molecule with the solid surface in the small channel. This packing

would lead to a smaller -AH and would suggest assigning Process 2 to the small channel.

The methane enthalpy assigned to the larger channel (Process 1) is that expected if

methane adsorbs with the face of the tetrahedron on the solid. With this assignment

methane falls on both the large channel and small channel enthalpy-entropy plots.

Adsorption of ethane, like propane and SF6, resulted in a reversal of the enthalpy-

entropy assignments. Examination of the entropic term for the second process of ethane

indicates a more negative contribution than propane or SF6. If one included ethane on the

above enthalpy-entropy plot, Figure 2-9, the corresponding point would lie well above the

best fit line indicating a much more ordered adsorption for ethane. This can be explained

by the size of the ethane molecule in relation to the channel diameter. ZINDO molecular

modelling of ethane indicates that the maximum length of the molecule is 4.821 A. The

dimensions of the smaller channel are reported as 5.1 x 5.4 A. Of the molecules examined,

ethane is the only adsorbate capable of spanning the smaller channel, perpindicular to the

surface, and interacting with both walls. Ethane adsorption in the larger channel is

believed to occur with the bond axis of the molecule parallel to the surface in an eclipsed

conformation. Further support for this will be offered in an analysis of the MEA surface

areas and pore volumes, vida infra.

Dimethyl ether was not included in these correlations. Adding them to the

enthalpy-entropy plots indicates that specific interactions are involved in Process 1 and

weak hydrogen bonding in Processes 2 and 3.

Further support for the above proposals arises from a van der Waals plot. For

carbon, the square root of the van der Waals, a, parameter plots up linearly with the

Process 1 enthalpies (R2 = 0.99). Figure 2-10 shows a similar plot for HZSM-5. For the

smaller channel, a fair correlation results (R2 = 0.96) indicating a general trend. A linear

van der Waals plot is expected only if the Type of interaction is the same for all molecules

involved. Thus, the slope of the plot shown has a contribution from changing the Type of

interaction for the small and large molecules. This behavior is contrasted to carbon where

the adsorptive selects pores from the distribution available to undergo similar adsorbate-

wall interactions for all molecules involved in the iV processes. This plot supports the

proposals offered above in the discussion of the enthalpy-entropy plots. The small channel

methane is lower than expected and the larger channel methane is higher.

A wide variety of probe molecules have been studied on carbonaceous adsorbents

and none show the reversal in enthalpy and equilibrium constant shown by HZSM-5

adsorption of ethane, propane and SF6. In contrast to HZSM-5, the pores of the

carbonaceous A-572 consist of parallel graphitic sheets or planes. The adsorbate-

adsorbent interactions occur along the surface of these planes, and a large adsorbate is less

constrained than it is in a cylinder. This suggests that the reversal in K and AH is more a

function of the cylindrical shape of the zeolite pores than the fixed pore size distribution.

Slit shaped pores do not require the extent of adsorbate reorganization that cylindrical

ones do.

14I I

12 *Small Channel

. 10 Large Channel




2 I
1.0 1.5 2.0 2.5 3.0

vdW [a]1/2

Figure 2-10: -AH vs. vdW [al"2 for HZSM-5

Adsorption by Silica Gel

Adsorption isotherms on Fisher silica gel, FSG, are also best fit by a two process

analysis of the adsorption data. Figure 2-11 shows the adsorption isotherms for the non-

condensible gases studied at -440C. The enthalpies and equilibrium constants for Process

1 with silica gel are smaller than those for HZSM-5, Table 2-7. Amorphous silica consists

of larger pores the HZSM-5 and thus lacks the concentrating power found with smaller

pores. These differences dramatically illustrate the influences of porosity on adsorption.

The propane enthalpies suggest that the pores of amorphous silica are larger than the

pores of HZSM-5. However, the N2, CO and CII enthalpies are comparable to those for

the larger channel in HZSM-5 indicating that either there are small pores in FSG that are

not accessed by propane or FSG has larger pores but a more polarizable surface.

Reversals in process enthalpies are observed for ethane and propane with silica gel.

Although silica is amorphous, it is comprised of cylindrical pores which, as was observed

for HZSM-5, invoke constraining effects on larger adsorbate molecules. This reversal in

an amorphous solid supports our assignment of reversals to the channel as opposed to the

slit nature of the pores. Thus, cylindrical pores exist in both amorphous silica and HZSM-

5. The entropy-enthalpy plot for the small pores is given in Figure 2-9. For the

adsorbates studied with silica, the enthalpies and entropies for Process 1 of N2, CO, and

CH4 and the enthalpies for Process 2 of ethane and propane give a fair correlation. The

magnitudes of the small pore enthalpies also plot up linearly with the square root of the

0 .0005 I I I I I I i | I I I I I I I I I
-440C Adsorption Data

0.0004 CH4
S~. CO

W 0.0003 2 N

S 0.0002


0 0.2 0.4 0.6 0.8 1.0


Figure 2-11: Adsorption Isotherms for the Non-Condensible Gases Studied with FSG

Table 2-7: Equilibrium Constants, n-values, Enthalpies, and Free Energies

N2 / FSG
n, = 0.21310.001 n2 = 2.480.07
-AH1 = 4.20.4 -AH2 = 2.910.1
TempC Ki -AG1 K2 -AG2
-94 4.910.4 0.567 0.290.03 -0.436
-62 0.90.4 -0.015 0.0910.03 -1.014
-44 0.50.2 -0.275 0.050.02 -1.367
-17 0.140.09 -1.015 0.030.01 -1.822

nj = 0.45110.002 n2 = 2.20.1
-AHI =4.1+0.2 -AH2 = 3.60.2
TempC KI -AGi K2 -AG2
-93 71 0.710 0.610.2 -0.175
-62 1.40.3 0.149 0.140.05 -0.827
-44 0.70.1 -0.147 0.060.02 -1.265
-17 0.230.09 -0.739 0.030.02 -1.747

nl = 0.330.001 n2 = 3.030.14
-AHI =4.610.1 -AH2 = 3.20.1
TempTC Ki -AG1 K2 -AG2
-93 20.12.5 1.073 0.60.2 -0.185
-62 2.80.3 0.437 0.160.03 -0.765
-44 1.30.3 0.105 0.090.03 -1.074
-17 0.450.2 -0.408 0.040.02 -1.649
ni in mmoles, Ki in atm", -AHi in kcal mole'

Table 2-7 continued

C3 / FSG
nl = 0.1110.0001 n2 = 2.80.2
-AHI = 4.90.8 -AH2 = 6.80.8
TemprC K1 -AGI K2 -AG2
75 2.90.5 0.731 0.10.02 -1.599
125 1.020.5 0.0156 0.030.02 -2.686
150 0.90.2 -0.1248 0.020.01 -3.444

n, = 0.2840.002 n2 = 2.30.2
-AH = 5.00.4 -AH2 = 5.60.3
TempC KI -AGI K2 -AG2
40 0.70.3 -0.022 0.080.04 -0.157
55 0.50.2 -0.045 0.050.02 -0.195
75 0.330.1 -0.083 0.030.01 -0.242
100 0.190.07 -0.123 0.0180.008 -0.298
ni in mmoles, Ki in atmn', -AHi in kcal mole'

van der Waals constant with the process reversal assignment, Equations 2-14 and 2-15.

The small range in enthalpies and entropies for the larger pores of FSG precludes an

analysis of the enthalpy entropy relations or the van der Waals plot.

-AHI = 1.458667 (0.09) a'1 + 2.380663 (0.19) R2 = 0.98 Equation 2-14

-AH2 = 1.163552 ( 0.3) a"2 + 1.782741 ( 0.57) R2 = 0.86 Equation 2-15

Utility of the Equilibrium Parameters

The multiple equilibrium parameters provide unprecedented detail about the

adsorption process. The calculated capacities of the processes, nj,, in moles and the

equilibrium constants, KId,, can be substituted into Equation 2-3 as was illustrated for

carbon, to resolve the total isotherm into the Process 1 and 2 components. This resolution

is shown in Figure 2-12 for CH4 with HZSM-5 and shows that adsorption occurs

simultaneously in all pores of the solid during the adsorption process.

Of particular importance are the surface areas and pore volumes for the individual

processes that can be calculated from the n's of the MEA using the molar volume and

cross sectional area of the adsorbate as was done for carbon in the previous section.

Adsorbate volumes and areas vary considerably in different literature reports and as such

were calculated using ZINDO molecular modelling.4 The molecular areas and volumes

used in the determination of MEA surface areas and volumes are listed in Table 2-8. The

calculated MEA accessible surface areas reported in Table 2-9 for HZSM-5 are N2, 269

m2g'; CO, 382 m2g'; CH4, 363 m2g'; CH6, 367 m2g'; C3Hg, 357 m2g' and SF6, 409

m2g''. The surface areas calculated are based on the maximum dimensions of the

molecule which best represents the bulk area covered on the surface of the solid per

molecule. The selected molecules will pack differently due to variations in their electronic

structure so that the tetrahedral or octahedral complexes of CH4 and SF6 may pack tighter

than the linear molecules. N2 has the lowest surface area of the probes examined which is

attributed to the strength of interaction between the gas and the solid. Methane and the

remaining adsorbates are more polarizable which affords a greater surface interaction and

the possibility for more adsorbate to be adsorbed.


o 0.0015



0 0.2 0.4 0.6 0.8 1.0

Pressure [atm]

Figure 2-12: Process Resolved Isotherms for CH4 Adsorption by HZSM-5

Ethane, as discussed earlier, is believed to span the diameter of the smaller channel

perpendicular to the surface, and parallel the surface in the larger channel. In the smaller

channel, ethane would have three hydrogens interacting with the solid surface on both

walls. The molecular surface area for ethane in this configuration is 20.509 A2, as

calculated by ZINDO. For the larger channel, ethane is expected to lay flat, in an eclipsed

conformation, with four hydrogens interacting with the surface. The molecular surface

area for ethane in this configuration is 25.325 A2. The calculated MEA surface

Table 2-8: ZINDO Calculated Molecular Surface Areas and Volumes
for Selected Probes

Adsorbed xbyy xbyy xbyz xbyz xbyz xbyz xbyz
xX 2.991 4.279 4.101 4.079 6.602 5.267 6.311
y., 3.054 3.339 3.942 4.821 4.554 4.872 4.166
z., 4.046 3.183 3.829 3.809 4.074 5.555 4.285
Occupied Surface 15.974 17.978 20.728 20.509 35.503 38.621 35.696
Area [A] 25.325
Calculated Molar 25.167 28.444 29.907 47.472 64.578 77.524 56.683
Volume [mL mol1] 47.477

area for ethane adsorbing in the two channels as described is 367 m2 g"' which is in very

close agreement with the MEA surface area of methane. Correlation of this information

with that presented in the enthalpy-entropy plots adds further support for MEA to

distinguish pores of different dimensions and provide an accurate assessment of porosity

for porous materials. The increase in the surface area for SF6 is attributed to its ability to

overlap into the channel intersections and interact with the available surface.

The values of the surface area from MEA are only slightly lower than the value of

402 m2 g' from BET. The BET model is generally applicable to the determination of

surface areas for mesoporous and macroporous solids where multilayer adsorption is

possible. However, for solids with micropores, the BET is not applicable, not only for

those reasons introduced in Chapter 1, but also that, in microporous materials,

condensation (multilayer formation) of an adsorbate is not possible. In the absence of

multilayer formation, the BET model simplifies to the Langmuir equation.3 The MEA

model is therefore applicable to the determination of surface areas for microporous

materials, however, unlike Langmuir or BET, MEA is capable of resolving subtle

differences in adsorption equilibria which, as for ethane, provides insight into the ordering

of adsorbate molecules.

As expected, the process capacities in mmoles g' (Tables 2-5 and 2-9) decrease as

the kinetic diameters of the gas molecules increase from 3.64 to 5.5 A The reported total

H-J t-plot pore volume for HZSM5 at 77K with N2 was found to be 0.15 ml g'.

Calculation of the total pore volume by the MEA (Table 2-9) for N2 indicates a total

volume of 0.07 ml g'. The other adsorptives gave values closer to 0.1 ml g-. Clearly

MEA is calculating pore volumes and surface areas for HZSM-5 which are in close

agreement with accepted models for determining micropore volumes. At the temperatures

ofporosimetry measurements, the channel intersections are expected to be filled with

Table 2-9: Resolved Process Capacities and Pore Volumes

N, N7
Process mmol ads ml ads area[mg] mmol ads ml ads area[mg ]
1 1.542 0.038 144.25 0.213 0.0054 20.5
2 1.289 0.033 125.01 2.48 0.0624 238.5
Total 2.831 0.071 269.26 2.693 0.0678 259

Process mmol ads ml ads area[mg'] mmol ads ml ads area[m2g]
1 0.6089 0.0173 65.9 0.451 0.0128 48.8
2 1.22 0.0347 132.04 2.2 0.063 238.1
3 1.7 0.0484 183.99
Total 3.5289 0.1004 381.93 2.651 0.076 286.9

Process mmol ads ml ads area n'i] mmol ads ml ads area[m g]
1 1.99 0.0595 248.32 0.33 0.0098 41.5
2 0.92 0.0275 114.8 3.03 0.0906 378.1
Total 2.91 0.087 363.12 3.36 0.1004 419.6

CH_ CiH.
Process mmol ads ml ads area[m ] mmol ads ml ads area[m g
1 0.8189 0.053 175.02 0.111 0.0072 23.7
2 0.85 0.055 181.67 2.8 0.1808 598.4
Total 1.6689 0.108 356.69 2.911 0.188 622.1

C2H c CHn
Process mmol ads ml ads area[m2g'] mmol ads ml ads area[m g]
1 1.92 0.091 292.72 0.284 0.0134 43.3
2 0.6 0.0285 74.05 2.3 0.1092 350.6
Total 2.52 0.1195 366.8 2.584 0.1226 393.9

Process mmol ads ml ads area[mg ]
1 0.56 0.043 130.2
2 1.2 0.093 279
Total 1.76 0.136 409.2

adsorptive making the deviation in the conventional porosimetry and MEA values for

non-condensible gases even larger than those for surface area. However, analysis of the

BJH desorption data for HZSM-5 indicates the presence of mesoporosity. Although

HZSM-5 and other zeolites are reported microporous materials, literature reports state

that under conditions of drying, ion exchange, dealumination, and deammination textural

damage can develop giving rise to mesoporosity.4

Further analysis of the n values of HZSM-5 adds support for the assignment of the

processes previously discussed. Examination of the process volumes for Process 1 of N2,

Process 2 of CO, and Process 2 of CH4 indicate similar capacities, approximately 0.03 mL.

This supports the assignment of methane's Process 2 to the smaller channel of HZSM-5.

SF6 indicates a larger volume than do the other probes. This can be attributed to SF6

overlapping into the channel intersections. Based on the similarities of the total reported

surface areas for CH4, CO, ethane and propane, the MEA surface area for HZSM-5 is

reported as 366 m2 g-.

Likewise, examination of the volume capacities for silica indicate that all of the

gases share similar volumes, except for N2 and CO. The lower volume obtained for N2 is

attributed to its weak affinity for the surface of silica. The pore volume obtained from N2

porosimetry was 0.32 ml. This is larger than that calculated by MEA indicating

condensation in larger pores of the amorphous silica. The MEA surface area for silica

with methane, ethane, and propane is in close agreement with that of BET. This is

possible due to the BET calculation being performed in a pressure range of 0.05 to 0.3

torr, which would not allow for the condensation N2. The MEA surface area of silica is

reported to be 478 m2 g'.


The MEA model presented in this chapter has made an attempt to answer some of

the problems faced when interpreting adsorption equilibria. Porous materials, whether

amorphous or crystalline, contain a distribution of pore sizes and surface defects which

give rise to sites or processes of different adsorption potential. The MEA model provides

a method for separating these adsorption processes such that an understanding of these

multiple interactions can be obtained. The process capacities provide a means of

calculating accessible surface areas and pore volumes based on accepted values of the area

and volume of the adsorptive. Correlation of the temperature dependent equilibrium

constants provides the enthalpies of interaction for the processes. The adsorption

isotherm can be separated into the individual processes by substituting into Equation 2-3

the corresponding n and K values for a particular temperature set. Resolution of the

processes allows one to see at what pressures maximum capacity for a process is reached.

The carbonaceous adsorbents discussed presented a linear correlation of the

enthalpies, -AHi, with the square root of the van der Waals [a] parameter and

polarizability for the gases studied. This led to the belief that the interactions for the

different sized adsorbates were similar for Process 1 regardless of the probe. Such a

relation allows one to predict, on the basis of the known van der Waals parameter or

polarizability, the enthalpies or equilibrium constants for a new adsorptive. Clearly, this is

a step in the right direction for predicting adsorption isotherms for probe gases whose

interactions in the pores or channels would be assumed similar to those of previously

examined probes. For HZSM-5, correlation of the van der Waals parameter was not as

linear as the carbonaceous adsorbents. Examination of the adsorption interactions for the

different sized adsorbates revealed that linearity was close if the Type of interaction was

the same. The smaller adsorbates did not interact as strongly as the larger adsorbates

which approached the dimensions of the channels. Likewise, as the adsorbates

approached the dimensions of the channel, entropic contributions arose as a result of the

adsorbate undergoing a more ordered adsorption on the surface. These entropic

contributions resulted in lower equilibrium constants for those pores where the greatest

interaction (largest enthalpy) was occurring. These problems resulted in a Process, Type

designation for describing the interaction of the adsorbate molecule with the pores or

channels of the solid where the Process describes the ease of adsorption and the Type the

matchup of the probe and pore dimensions.

Comparison of MEA surface areas and pore volumes to N2 porosimetry indicates

that MEA surface areas are generally lower than traditional BET measurements for carbon

which may be a result of N2 condensing in the larger pores of the solid For HZSM-5, the

small pore channel dimensions do not allow for molecules to aggregate preventing

condensation which in turn reduces the BET equation to the Langmuir equation providing

a close correlation of the two models with MEA. Traditional N2 porosimetry

measurements at 77K do not adequately describe the adsorption process since normal

applications of these materials occur at or above ambient temperatures. Multiple


temperature analyses provide a better description and allow for a better correlation of the

adsorption data with known parameters of the solids and related adsorptives. Ratioing of

the areas of tested probes and related surface areas affords the prediction of process

capacities and surface areas for untested probes.

The MEA model is capable of presenting useful information in gas-solid equilibria

based on the general principles of thermodynamics. Many of the important properties

which are not easily obtained from current adsorption and surface area models which

should be considered in selecting appropriate adsorbent systems can be obtained from this

model based on the well defined physical properties of the adsorbates.



Porous materials including carbon, silica gel, and zeolites are used in many

practical applications with liquids including separation and purification as well as catalysts

supports. In order to understand the competitive binding involved in these applications, a

fundamental understanding of the relevant equilibria is necessary. An MEA analysis of

gas-solid equilibria for several probe gases examined with porous carbonaceous and silica

based adsorbents was discussed in Chapter 2. Adsorption isotherms were resolved into

three adsorption processes for the adsorbates studied with the carbonaceous adsorbent,

Ambersorb 572. In the absence of specific interactions, the first process corresponded to

adsorption of the probe into the smallest accessible pores of the solid, and the second

process to pores with larger diameters. Silica gel indicated the presence of two adsorption

processes for the adsorbates examined. In our study of gas-solid equilibria, dispersion

type forces dominated the adsorption of nonpolar gaseous probes by the carbonaceous and

silica based adsorbents. To extend the applications of the Multiple Equilibria Analysis

(MEA) model further,4 the competitive adsorption of liquid probes from dilute solution

with silica gel and the carbonaceous adsorbent Ambersorb572 (A-572) were investigated.

Adsorption from solution involves several different molecular interactions which

are shown in Figure 3-1. Al and A3 represent the affinity of the solid for adsorption of

solute and solvent respectively, and A2 is the interaction of the solute with the solvent. A2

is related to the solubility of the solute in the solvent and, for non-interacting systems, is

related to the difference in the solubility parameters. Both A, and A3 involve interactions

with the accessible surface area of the solid and are influenced by the adsorbate and

solvent nonspecific London dispersion interactions, dipole-dipole interactions, and donor-

acceptor interactions. In the case of a porous adsorbent with little donor-acceptor

functionality, A, and A3 are dominated by nonspecific London dispersion interactions for

nonpolar adsorbates and dipole induced dipole interactions for polar adsorbates which are

usually larger. These forces correlate typically to the enthalpy of vaporization (AHK) of

the adsorbate.


Porous A2

A3 Solvent

Figure 3-1. Model for Liquid Adsorption in Porous Solids

This work investigates the tendency ofA-572 and silica gel to preferentially adsorb

a donor solute from a dilute binary solution. The acidic sites on the surface of silica gel

lead to specific donor-acceptor interactions which are larger than the non-specific

interactions described above. Previously, the Cal-Ad47 method was able to elucidate three

hydrogen-bonding sites of different strengths when pyridine was used as the basic probe

molecule. In order to extend the applications of the MEA model and to investigate the

competitive interactions (A1, A2, and As) of an acceptor solid such as silica gel to

preferentially adsorb specific donor type molecules, several dilute binary systems have

been investigated and are reported here. The preferential adsorption has been interpreted

using the MEA model to obtain equilibrium constants (Ki, .) for adsorption to the surface

of silica gel and A-572.


Materials. Silica Gel, FSG, (Fisher S-679, lot # 934403) was obtained from

Fisher Scientific and Ambersorbe572 (A-572, lot# 2201) was obtained from Rohm and

Haas. Both solids were dried in a vacuum-oven at room temperature for 24 hours prior to

use. The FSG is a porous, amorphous solid with a surface area of ~570 m2g"' with an

average pore size of 28 A as determined from the N2 isotherm at 77K using a 5 point BET

calculation25 and the BJH desorption isotherm.35 FSG contains micropores (0.02 mL g')

and mesopores (0.33 mL g'l) as determined by nitrogen porosimetry. A-572 is a

carbonaceous adsorbent with a surface area of 1100 m2 g' and a total pore volume of 1.1

ml g', and is comprised of micro- and mesopores. Gas-solid measurements were

performed on a Micromeritics ASAP 2000 gas-solid analyzer. Probes and solvents were

cyclohexane (certified ACS spectroanalyzed), pyridine, benzonitrile, nitrobenzene, p-

xylene, benzene and chlorobenzene.

UV/VIS spectroscopy. All UV/VIS spectra were measured using either 1.0 or

0.1 cm quartz UV cells in a Perkin-Elmer Lambda 6 UV/VIS spectrophotometer.

Adsorption measurements. Approximately 0.22 to 0.23 g of silica or A-572 was

precisely measured into 12 vials. A 1.00 mL aliquot of known concentration of probe in

solvent spanning a desired concentration range was then pipetted over the solid. The vials

were sealed and wrapped with parafilm, and the samples were allowed to equilibrate for

16-24 hours with occasional agitation. The equilibrium concentration of the probe in

solution was determined from calibration curves of absorbance versus concentration. In

order to correct for errors incurred from large dilutions, absorbance measurements were

made from two or more separate dilutions and an average value recorded.

The total number of moles of probe adsorbed per gram of solid adsorbent, n, was

calculated using Equation 3-1.

(C, C,)V(0.00L / mL)
n= Equation 3-1

where Ci is the initial probe concentration (mol L'), C, is the equilibrium probe

concentration (mol LU), V is the volume of solution added (mL), and m is the mass of

adsorbent (g). Total adsorption isotherms were obtained by plotting n vs. C, for each

sample vial.

Equilibrium analyses. The measured isotherms were analyzed with a multiple

process equilibrium analysis model for porous solids which was described in Chapter 2.

The equilibrium expression derived in earlier studies of gas-solid equilibria is shown in

Equation (3-2) where SA is the total number of moles of solute adsorbed per gram of

solid, ni, is the capacity of process i in moles, KY .is the equilibrium constant of

adsorption for process i, and C, is the equilibrium solute concentration in solution. The

two parameters, ni and K.s, are calculated using a modified simplex routine designed to

fit measured equilibrium concentrations and [SA]for a series of probe concentrations to

multiple equilibrium equations. The number of processes used in the fit was either 1 or 2

depending on the minimum needed to obtain a good data fit.

[ = niK[Ceq]
[SA] = K Equation 3-2

Results and Discussion

Adsorption from cycloheane. In this work, various probe molecules were

studied in the same solvent. Cyclohexane was selected as the solvent since it is a non-

polar, poorly solvating solvent. These properties minimize the solute-solvent interactions

(A2). Using the same solvent keeps A3 constant for all of the systems investigated. The

probe liquids used in the adsorption measurements were selected to encompass a range of

Table 3-1: Summary of Solutes and Cyclohexane and Physical Properties4

Probe MW Polarizabi van der Dipole Molar S" Ebb C, AH,
Liquid [g/mol] lity Waals Moment Volume [kcal/mol]
[A 3 [a] [Debye] [mL/mol]
Pyridine 79.10 2.22 80.9 2.44 1.78 3.54 8.39
Benzonitrile 103.12 12.5 33.39 4.18 102.1 2.63 1.65 0.75 11.0
Nitrobenzene 123.11 12.9 4.22 102.9 2.61 1.27 0.57 13.2
Benzene 78.11 10.3 18.00 0 89.4 1.73 0.70 0.45 7.34
p-Xylene 106.17 14.1 30.93 0 122.6 --- --- --- 8.60
Chlorobenzene 112.56 12.3 25.43 1.69 101.7 2.07 --- -- 8.73
Cyclohexane 84.16 11.0 22.81 0 108.0 1.11 0 0 7.16

electrostatic and covalent donor properties. In order to study the possibility of x-

interactions of the weak donor probes with the surface, various substituents on the

benzene ring were used to influence the basicity of these interactions. Properties of these

materials are summarized in Table 3-1.

Adsorption isotherms. Figure 3-2 shows the adsorption isotherms for various

solute molecules adsorbed from cyclohexane solvent by silica gel. Except for benzene and

chlorobenzene, all of the solutes studied produced non-linear isotherms which required

two processes to obtain a data fit that was within the accuracy of the measurements. One

process was sufficient to describe the adsorption of benzene and chlorobenzene. The

points in Figure 3-2 correspond to the experimental data, and the curves drawn through

the data are generated from the best fit equilibrium analysis. The experimental data points

and the calculated values for each point are presented in Appendix E.

Table 3-2 summarizes the n and Ki, parameters for all of the isotherms where n is

the capacity of the solid for the adsorption process in number of millimoles per gram of

solid, and Ki, is the adsorption equilibrium constant which is a measure of the affinity of

the probe for the solid. Since silica gel is an acidic solid, KI,. for all of the adsorbates

refers to the affinity of the probe for the strongest adsorption sites. Kz2. refers to the

potential for multilayer adsorption of the adsorbate and adsorption to weaker surface sites.

It should be emphasized that the KiQ. values correspond to an average value of different

processes that have K. values close enough to be treated as a single process in the data

work up within experimental error of the measurement. Vi and V2 correspond to the

volume of solute adsorbed via process 1 and 2. These are obtained from the product of

0.0030 1 I 1 I 1 1 I
---- Pyridine
0.0025 1-- ..---* CN-benzene
S--- NO -benzene
|o 0.0020 -- p-Xylene

E-- e- Benzene
; 0.0015 .... -... Cl-benzene
-0 -
S0.0010 .

S 0.25 0.50 0.75 1.00 1.25 .50


Figure 3-2: Adsorption Isotherms for Selected Adsorbates with Silica Gel.

Table 3-2: Adsorption Parameters for Adsorption from Cyclohexane by Silica Gel.

Probe K,.&s ni Vi,. K2d, n2 V2,.
[_____mmol] _[m] [mmol] [ mL
Pyridine 577+159 1 S+0 6 0.128 2+149 1 67+10 6 0.135
p-Xylene 23510 0.120.04 0.015 2.20.8 1.30.5 0.162
Benzonitrile 22450 0.90.2 0.092 637 17 0.121
Nitrobenzene 10819 0.70.1 0.072 59 12 0.152
Chlorobenzene 2.50.2 0.850.08 0.086
Benzene 0.70.3 2.30.8 0.206

the molar volume (mL mol1) of the solute and the calculated capacities, n, expressed in


Considering the adsorption of the various probes on the surface of silica gel,

pyridine has the greatest affinity and benzene the smallest affinity for the strongest sites on

the surface. The ordering of Kil, is readily explained by the donor strength of the

adsorbate and the resulting solute-solid interactions. When silica is the solid, A, is

influenced by donor-acceptor interactions between the probe molecule and the surface. Of

the probes examined, pyridine was the most basic and is represented by the largest Kti&.

P-Xylene experienced the second highest Kl,., followed by benzonitrile, nitrobenzene,

chlorobenzene and finally benzene. Of these probes, pyridine, benzonitrile and

nitrobenzene normally react via the basic lone pairs of electrons. The basicity of these

lone pairs determines their ability to undergo acid-base interactions. Benzene,

chlorobenzene and p-xylene normally hydrogen bond through the pi system of the

aromatic ring. The order of the equilibrium constants for Ki indicates that, as expected,

the electron donating ability of the methyl groups of p-xylene increases the basicity of the

aromatic pi system over that of benzene and chlorobenzene.49 Nitrobenzene interacts with

the acidic sites of silica through the nitro group. The interaction of nitrobenzene is weaker

than that ofbenzonitrile. Interaction through the pi system is not expected to occur as the

basicity of the pi system for nitrobenzene is reportedly less than that of the nitro group.9

Chlorobenzene has a slightly greater equilibrium constant than benzene. The lone pairs of

electrons on chlorine may contribute to the basicity of chlorobenzene and allow for weak

hydrogen bonding interactions with the acidic sites of silica. The pi donor, benzene

exhibited the weakest adsorption affinity of the probes examined.

As was described in Chapter 2, the van der Waals [a] parameter describes the

relative attraction between like molecules. A plot of the square root of this parameter

with the In Ki should give a slope corresponding to the relative attraction between the

probe and the solid. However, such a plot is valid only for those adsorbates undergoing

non-specific adsorption interactions with the adsorbent. In minimizing the contributions of

A3 shown in Figure 3-1, the adsorption of the selected adsorbates was performed in the

same solvent such that the adsorbates experience the same type of competitive adsorption.

Subtraction of the van der Waals parameter for the solvent from the vdw [a] parameters of

the probes has been suggested, however this only subtracts a constant and does not affect

the parameter which is more greatly affected, Ki. Correlation of the In Ki with square

root of the van der Waals [a] parameter for benzonitrile, benzene, p-xylene and

chlorobenzene resulted in an R2 of 0.87. This is shown in Figure 3-3. The increasing

order of Ki for silica does not follow the order of increasing van der Waals parameter of

the probe molecules as was observed for the gaseous adsorbates previously reported.

Therefore the correlation of the van der Waals [a] parameter with In K,.. is not

appropriate for predicting equilibrium constants for the interaction of new donor

adsorbates with silica.







-1 1 ,
4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8

Figure 3-3: In K, vs. vdWIa]1' for benzonitrile,
benzene, p-xylene and chlorobenzene.

The E and C model has been developed to describe the electrostatic and covalent

contributions of donors and acceptors in acid-base interactions.50 Although this is a very

limited data set, an ECW analysis of silica with the adsorbates whose EB and CB

parameters are known resulted in an EA of 6.37 ( 1.78) and CA of-0.20 ( 0.57) with W

= -4.34 ( 2.1), and an R2 of 0.95. The equation is shown below, Equation 3-3. The E

and C model is a dual parameter model; however, within experimental error, CA for silica

is zero. Therefore, electrostatic forces are dominating the interactions of the selected

adsorbates with silica.

In K = 6.37 ( 1.78) EB 0.20 ( 0.57) CB 4.34 ( 2.1) Equation 3-3

The Ki,, values characterize the adsorptive properties of different solids and different

adsorbates. Since In Kl,. increases with increasing EB for those adsorbates whose EB

values are known, (Table 3-1) the affinity of a solid for different adsorbates may be

predicted from the electrostatic component of the adsorbate donor character. Correlation

of the EB parameters with In KI for pyridine, benzonitrile, nitrobenzene and benzene

resulted in an R2=0.94 and is shown in Figure 3-4.

Examination of the equilibrium constants for the second adsorption process, K2,,

with EB of pyridine, benzonitrile, nitobenzene and benzene resulted in a poor correlation.

However, examination of in K2. with the heat of vaporization for pyridine, p-xylene,

benzonitrile and nitrobenzene indicates a linear relationship. A least squares analysis of In

K2d with the -AHy,, resulted in an R2 of 0.95. Therefore, after the adsorbates have

interacted with strongest acid sites of silica, the remaining adsorption corresponds to

multilayer formation in the pores of the solid.

In contrast to the adsorption parameters observed for silica, Ambersorbo 572

experiences slightly different interactions with the probes examined. A-572 is an

amorphous solid whose structure, like other carbonaceous adsorbents, has been



-1 I i-----

0.6 0.8 1 1.2 1.4 1.6 1.8

Figure 3-4: Relationship of In K.,.., with EB of the selected adsorbates studied with

described as an extended array of aromatic rings comprising a slit-shaped pore geometry.

Table 3-3 lists the adsorption equilibrium parameters and Figure 3-5 presents the

adsorption isotherms for the liquid adsorptives examined with A-572. From Table 3-3 it is

clear that the affinities of the selected adsorbates are not the same for A-572 as were

found for silica. Two adsorption processes were found to best define the adsorption

isotherms for the probes selected, except for benzene. Adsorption of pyridine by A-572

was characterized by a weaker Ki than for silica indicating the absence of any strong acid

sites. Benzonitrile and chlorobenzene adsorption exhibited the strongest affinity for

adsorption, more so than with silica, while benzene exhibited the weakest. An E and C

analysis ofA-572 with the probes listed in Table 3-3 is given in Equation 3-4.

Table 3-3: Adsorption Parameters for A-572
Probe Ki,.d nI VIdI KIId. n2 V,!,
I[mmol] [mL] [mmol] [mL]
Benzonitrile 635012 0.501 0.050 323 170.3 0.174
Chlorobenzene 11028 0.430.04 0.044 83 1.30.4 0.131
Nitrobenzene 31415 1.10.1 0.110 147 21 0.206
Pyridine 300260 0.50.4 0.041 3+38 34 0.245
Benzene 0.81 69 0.513

In K = 9.94 (0.18) 1.58 (0.06) -6.40 (0.21) R2= 0.999 Equation 3-4

For A-572, electrostatic interactions are found to be stronger. Clearly, the strength of the

interactions involved in the adsorption of the selected adsorbates does not follow that of

their basicity, or EB parameter. A least squares analysis of In Ki with EB for benzonitrile,

nitrobenzene, pyridine and benzene resulted in an R2 of 0.77.

Examination of the square root of the van der Waals [a] parameter with In Ki for

benzonitrile, benzene and chlorobenzene resulted in an R2 of 0.9. If one examines Table

3-1, the increasing values for the van der Waals [a] parameter follows the increasing order

of the K, values reported in Table 3-2, except for benzene. In the study of adsorption of

0.5 1.0 1.5 2.0 2.5

Ceq ]

Figure 3-5: Adsorption Isotherms for Selected Adsorbates with A-572.




G 0.001


these probes from cyclohexane, all of the probes except benzene have a greater van der

Waals [a] parameter than the solvent. This would indicate a stronger attraction for

cyclohexane by A-572 thereby making it difficult for benzene to displace the solvent. This

indicates that for the carbonaceous adsorbent, nonspecific dispersion forces are

dominating the adsorption process.

Examination of the equilibrium parameters for the second adsorption process ofA-

572 indicates slightly larger adsorption interactions are occurring in carbon than were

observed for silica. Examination of In KZd. with the enthalpy of vaporization of the

adsorbates results in a poor correlation (R2 = 0.4). Therefore, the interactions of the

second process relate to adsorption of the probe molecules in pores approximately 1.5

times the size of the adsorbate. This was illustrated in Chapter 2 where the multiple

adsorption processes observed for A-572 with gaseous adsorbates resulted in a set of

equilibrium constants of which the largest equilibrium constant, Ki, referred to adsorption

into the smallest accessible pores which closely matched the size of the adsorbate. The

second largest equilibrium constant referred to adsorption of the gaseous adsorbates into

pores which were approximately 1.5 times the size of the adsorbate, and the smallest K.&

referred to pores which were roughly twice the size of probe. The adsorption of the

gaseous adsorbates, as well as the liquid adsorbates studied here, have increasing

equilibrium constants which relate linearly to an increase in the van der Waals [a]

parameter for the corresponding probes. Although this data set is very limited, a general

trend can be established for the prediction of the affinity of new adsorbates with A-572

based on the known van der Waals constants of the new probe.

The n values calculated for the adsorption processes relate to the number of acid

sites available or accessible to the various probe molecules. For silica gel, pyridine had the

largest ni and is interacting with the strong and weak sites of the solid. Benzonitrile,

nitrobenzene, and chlorobenzene share similar capacities, and volumes for the first process

which may involve the interaction of these probes with the strong sites of the solid. P-

xylene is slightly bulkier with respect to the methyl groups which may prevent its

accessibility to smaller pores that may contain the majority of the stronger acid sites thus

contributing to its low capacity and pore volume for process 1. Within the experimental

error of those probes having a second process, the volume adsorbed corresponded to

approximately 0.15 mL. In comparison, the volume capacity of silica for the adsorbates

examined in Table 3-2 is larger than that found in gas adsorption studies where the

average volume capacity was 0.06 mL g'. This indicates the presence of multilayer

adsorption on the surface of silica.

Analysis of n for A-572 (Table 3-3) indicates that pyridine, benzonitrile and

chlorobenzene share similar capacities. The capacity of nitrobenzene is approximately

twice that of the other probes. Due to the reported slit-shape pore geometry of

carbonaceous adsorbents, chlorobenzene, benzonitrile and pyridine would have easier

access to smaller pores, whereas nitrobenzene may encounter difficulty with the nitro

group forcing it to seek larger pores. Benzene, whose van der Waals attraction parameter

is less than the solvent, encountered difficulty in displacing cyclohexane and resulted in

large errors in its calculated capacity. Likewise, comparison of the volume capacity for

gaseous adsorbates examined with A-572 indicates a striking similarity to the volume

capacity of the liquid adsorbates studied here. This indicates that in the liquid or gaseous

state, adsorbate molecules are accessing the same pores.


The affinity of various donor molecules for adsorption from cyclohexane solvent

by silica gel and a carbonaceous adsorbent has been compared using a MEA interpretation

of the equilibrium data. The K.d values obtained from an MEA fit give an indication of the

donor-acceptor properties of a solid which can be used to one's advantage in selecting

adsorbents to purify liquid mixtures. This information will serve to provide an indication

as to the potential purity that can be achieved in the separation of a mixture. If K.a values

of two materials for a solid are comparable, it would be difficult to get an acceptably pure

product stream. However, two materials with K.& values which are orders of magnitude

apart should provide a very good separation. Furthermore, this information is useful in

catalytic applications in which a support should be chosen which will have a preference for

adsorbing a reactant over a product. This would lead to a greater reactivity since the solid

would continuously adsorb reactant molecules on the catalyst surface and displace product

molecules. For silica based adsorbents, correlation of the EB and Ca parameters, if known,

for new donor probe molecules can provide an estimate of the adsorption equilibrium

constant for that new probe. For carbonaceous adsorbents whose interactions are

dominated by non-specific, dispersion type forces, prediction of adsorption affinities for

new probes can be accomplished by the van der Waals [a] parameter. However, care

should be taken when selecting solvents for the analysis of liquid adsorption if


carbonaceous adsorbents are the solids of interest so that competition between the solvent

and probe can be minimized.

Clearly, more probe molecules are needed to provide a better assessment of the

adsorption properties of these selected adsorbents and adsorbates by the MEA model.

Such a limited data set can only provide an indication of the adsorption potential of these

adsorbents for donor molecules. Further studies of liquid-solid adsorption equilibria by

the MEA model are currently underway.



Liquid acids, such as H2S04 and HF, have been used extensively for hydrocarbon

cracking, isomerization, and alkylation reactions in the production of petroleum and

specialty chemicals.51"s' The need for the products of these reactions results in many tons

of liquid acid waste being generated each year. Storage and transportation of acid and

acid waste is costly and has fueled debate over environmental and safety concerns in their


A solid acid capable of performing the same function as a liquid acid, which

provides an ease of product isolation and is stable and regenerable at moderate to high

temperatures has many environmental, as well as financial benefits.57 Solid acids generally

consist of a functionalized support matrix which contains an active complex that enhances

the acidity of the support. The active complex can be of a Bronsted or Lewis nature."

Zeolites, aluminosilicates, are solid acids which are used in many petrochemical

applications.13'16 The acidity of zeolites is a result of Lewis acid centers generated from

the incorporation of aluminum (III) in the tetrahedral matrix of silica (IV)."'"5 The

aluminum is tetrahedrally coordinated in the silica matrix generating a negative charge

which is balanced by a cation. Unlike many other metal oxide systems, zeolites are

crystalline solids which have defined pore and channel sizes. This benefits those reactions

where size exclusion is important. Zeolites occur naturally, but many of the more useful

types are synthesized in the laboratory which is costly, in both time and in money.'

Silica is one of the most widely studied materials for use as a catalyst support. 2

Silica gel itself is an acid whose acidity, although weak, is derived from the hydrogen

bonding of neighboring silanol groups which comprise the surface of the solid. The

density of these hydrogen bonding sites give rise to weak, moderately strong, and strong

acid sites. Chronister and Drago47 identified these acid sites in a Cal-Ad analysis of a

commercial silica gel, Figure 4-1.

H & H H H H

SSi Si Si- Si Si Si
/ 0 0O \

Figure 4-1: The Strong (left), Moderately Strong (Center) and
Weak (right) Acid Sites of Silica Gel

The strongest acid site was found to have a calorimetric heat of interaction of 12 kcal

mole"' with pyridine.

Enhancement of the acidity of silica gel was described by Getty and Drago who,

by reacting AIC13 with silica gel in CCl, found that a complex of the type shown in Figure

4-2 could be produced.'" A calorimetric investigation of the solid resulted in a heat of

interaction of- 50 kcal mole" with pyridine.


0 0

Si Si

Figure 4-2: Bronsted Acidity Generated from the AICi3 / Silica Gel Catalyst

The -AIC2 is a Lewis acid drawing electron density away from a silanol OH bond resulting

in a weakening of this bond, thus increasing the acidity of the proton. The result is the

generation of a strong Bronsted acid site. Although a large increase in acidity was

achieved, exposure to moisture resulted in deactivation of the solid.

Kob attempted to replace the sensitive -AiC12 with a sulfated W03 complex."9

Although effective, difficulties with migration and clumping of the WO3 on the surface

coupled with the loss of sulfate, and the difficulty with regeneration resulted in a loss of

catalyst activity after each use. Incorporation of an active complex into the matrix of the

solid to minimize migration and loss of the active complex was deemed necessary.

The sol-gel method is a convenient and effective procedure for incorporating metal

oxides into a silica matrix. The sol-gel procedure involves the hydrolysis, condensation,

and polymerization of a silicon alkoxide to yield silica gel."66 The most common silicon

alkoxide used is tetraethyl orthosilicate, TEOS. A standard silica sol-gel synthesis

requires approximately five steps. Hydrolysis, Step 1, ofa silicon alkoxide involves the

exchange of the alkoxy groups for hydroxyl groups resulting in the formation of silicic

acid. Alcohol solvents are generally used to aid in the miscibility of the TEOS and

water."6667 The alcohol is usually similar to the alkoxy groups being displaced, although

mixed alcohol systems are possible.

Si (OEt)4 + 4 H20 + EtOH + acid / base catalyst -+ Si (OH)4 + 4 EtOH (Step 1)

The hydrolysis reaction can be either acid or base catalyzed."'67 The choice will have an

impact on the nature of the porosity and surface area of the solid. Acid catalyzed

reactions generally result in small pore, high surface area materials, whereas base catalyzed

reactions develop an excess negative charge in solution resulting in larger pore, lower

surface area solids.'2

Condensation, Step 2, of the silicic acid results in branching and polymerization

with the formation of water.

Si (OH)4 + Si (OH)4 (HO)3 Si O Si (OH)3 + H20 (Step 2)

When the reaction species have polymerized, the free flowing liquid becomes a viscous,

non-flowing gel, Step 3. The gel is aged, Step 4, for a period of time at a controlled

temperature to ensure complete hydrolysis of the alkoxy groups. Most sol-gel reactions

are carried out in sealed vessels under autogenaceous pressure. If one desires, a

monolithic solid68 can be obtained by using a vessel lined with a material which has a low

affinity for wetting, such as teflon. This minimizes the stresses that develop due to

adhesion to the vessel surface if pyrex is used. After aging, the gel is dried, Step 5, in a

calcination oven at temperatures ranging from 300C to over 5000C with flowing air or

oxygen to remove any residual organic material.

One advantage of sol-gel derived silicas over those produced by commercial means

is that the syntheses are carried out at, or near, ambient temperatures and that relatively

pure solids can be obtained." Another advantage is the ability to incorporate additional

metal-oxo complexes within the silica matrix. By incorporating these complexes into the

structure, one is capable of controlling the relative acidity, catalytic activity, and physical

properties of the solid. The control of porosity and surface area is important for effective

catalytic and adsorption systems. Solids consisting mostly of large pores generally have

low surface areas, whereas smaller pores increase the overall surface area and

concentrating power of the solid. This is useful for systems where reactants must be

concentrated around a catalyst. For silica gel systems, increasing the surface area is also a

means of generating more acid sites on the solid. To control the pore size of the solid,

drying control chemical additives, DCCAs, can be used.670"7 These additives are

generally organic bases, such as formamide, which undergo hydrogen bonding with the

forming silanols of the reaction. A more uniform pore size can be established by evenly

dispersing the reagents throughout the reaction mixture. Once the solid is obtained, the

drying control additive is removed by solvent washing or calcination.

The incorporation of organic spacers has also been used to control the porosity of

solids." 75 Through a modified Grignard experiment, organic compounds such as phenyl

and biphenyl rings are placed between silicon atoms, Figure 4-3. This yields a route to

controlled pore size with the possibility of adding functional groups which may further
enhance the reactivity of the solid.

0- Si
0- Si






Figure 4-3: Organically Bridged Silica Gels

Mixed metal oxides, similar to those presented by Kob, can be synthesized by the
sol-gel route.7678 By incorporating these complexes into the silica matrix, the problems
observed with the tungsten catalysts may be avoided. Preliminary results for a sulfated
silica gel synthesized by the sol-gel method which has the potential of extended MTBE
production is described here.


Standard Sol-Gel

In most cases an 8:1:3 ratio ofH20, TEOS, and ethanol was used. A water to

silicon ratio of 8:1 is reported to be adequate to obtain the shortest gelation time and to

ensure complete hydrolysis of the alkoxide ligands.6 To 30ml of TEOS (98%, Aldrich

Chemical) was added 23 ml of 95% ethanol (AAPER Alcohol and Chemical, Co.), and 1

ml of concentrated HNO3 (Fisher Scientific). The reaction mixture was allowed to stir in a

Pyrex crystallizing dish, or a teflon cup, for 10 minutes before the addition of 18 ml of de-

ionized water. The mixture was stirred an additional 15 minutes before being placed in an

isothermal oven (80C) to gel and age. After a 24 hour drying period, the resulting solid

obtained is crushed and sieved through a 60 mesh screen to obtain a uniform particle size.

This is followed by calcination in air at 3500C to remove any residual organic material.

DCCA Sol-Gels

The use of drying controlled chemical additives, DCCAs,' 70-7' and organic

spacers in the synthesis of sol-gel silicas to control the porosity and surface area was

performed as follows: To 30 ml of TEOS (Aldrich) was added 23 ml of 95% ethanol, 1

ml of concentrated HNO3 (Fisher Scientific), and 10 ml of formamide (Fisher Scientific).

The reaction mixture was stirred for 10 minutes followed by the addition of 18 ml of de-

ionized water. Stirring was continued for another 15 minutes before being placed in an

isothermal oven at 80C. After a 24 hour period, the solid obtained was crushed and

sieved before calcination in air at 3500C to remove the drying control additive.

Phenyl and Biphenyl Bridged Silicas

1,4-bis(triethoxysilyl) benzene: 1,4-bis(triethoxysilyl) benzene was prepared as

described in the literature.7"75 To 15 g of magnesium metal turnings was added 450 mL

of TEOS, 300mL of THF, and 1 crystal of iodine. The reaction solution was then heated

to reflux. To this was added dropwise, over a two hour period, 48 g of 1,4 -

dibromobenzene (Aldrich) in 100mL of THF. This was allowed to continue refluxing for

an additional hour after the addition was complete before allowing to cool to room

temperature. THF was removed by vacuum distillation from the resulting grey-green

solution. Two hundred milliliters of n-hexane was added and the resulting solution was

filtered to remove the remaining magnesium metal. The n-hexane was removed by

vacuum distillation and the resulting oil was doubly distilled under vacuum to yield a clear,

colorless liquid. The resulting liquid is then combined with water, ethanol, and an acid

catalyst, as described above, to promote gelation and formation of the phenyl bridged


4, 4'-bis(triethoxysilyl) biphenyl: Preparation was reduced to 1/4 scale of the

reported literature synthesis.73"- To 25 g of 4, 4'-dibromobiphenyl (Aldrich) was added

200 mL of THF, 125 mL of TEOS, 10 g of magnesium metal, and one crystal of iodine.

The reaction solution was then heated to reflux under an argon atmosphere. After

cooling, the THF was removed by vacuum distillation and 200 mL ofn-hexane was added.

The solution was filtered under argon and the n-hexane removed by reduced pressure.

The remaining brownish-yellow oil was doubly distilled under vacuum to yield a clear,

colorless oil. The resulting solution was subjected to the afore mentioned sol-gel

preparation and the solid obtained characterized by N2 porosimetry and calorimetry.

Vapor Deposition of SO3

A Schlenk tube containing a 3 g sample ofpre-dried (100C, vacuum) sol-gel silica

was connected by a valved sidearm to a round bottom flask containing fuming sulfuric

acid (Aldrich Chemical). The Schlenk tube was degassed while the fuming sulfuric acid

was warmed to near boiling by a heating mantle. After evacuating the Schlenk tube for 30

minutes, the valve was opened and the acid vapor was allowed to equilibrate with the solid

for approximately 5 minutes. The valves were then closed and the solid sample was

evacuated and backfilled with N2 and moved to an inert atmosphere dry box.

Sol-Gel Derived Sulfated Silicas

In a teflon lined autoclave was added 30 ml TEOS, 23 ml EtOH, 15 ml SO2(OEt)2

(Aldrich), and Iml of concentrated H2SO4 (Fisher). The mixture was allowed to stir for

15 minutes before the addition of 18 ml of de-ionized water. Stirring was continued for an

additional 15 minutes before the autoclave was sealed and placed in an isothermal oven at

80C. The reaction was carried out for 3 days before the monolithic solid obtained was

dried. The solid was crushed, followed by calcination in air at 4000C for 12 hours. The

solid was allowed to cool for 3 hours open to the laboratory atmosphere before being

placed in a N2 atmosphere dry box. Substitution of SO2(OEt)2 by equal molar amounts of

H2S04 and NH4'SO42" was also performed. Samples consisting of 10% and 30%

SO2(OEt)2 were also synthesized.

Characterization of Solids

N2 Porosimetry

Characterization of commercial and sol-gel derived solids was performed on a

Micromeretics* ASAP 2000 gas-solid analyzer equipped with chemisorption and

physisorption software. Surface area and pore volume data was obtained from the N2

isotherm at 77K. Surface areas were determined from a 5 point Brunauer-Emmet-Teller,

BET, calculation.25 Micropore volumes were determined from the Harkins-Jura t-plot

model with thickness parameters from 5.5 -9.0 A34 The Barrett-Joyner-Halenda (BJH)

desorption curve was used for calculating meso- and macropore volumes.35 All

calculations were carried out using the ASAP 2000 software. Sample loadings were on

the order of 0.3 g and were dried at 200C under vacuum for a minimum of 8 hours prior

to study.

Calorimetric Analysis

In order to ascertain the solid's acidic strength, calorimetric titrations with pyridine

were performed and the results compared with previous solid acids. One gram samples of

commercial and sol-gel derived solids were titrated with 0.3M pyridine / cyclohexane

solutions. Cyclohexane was chosen because it is a poorly solvating and a poorly

coordinating solvent. The solids were dried under vacuum at 200C and stored in a N2 dry

box prior to examination. Cyclohexane, 100 ml, (Fisher reagent grade distilled over P20s)

was added to the calorimetric cell to create a slurry with the solid. Injection of the

pyridine solutions was performed using a Hamilton gas tight syringe with calibrated stops.

After each injection the heat of interaction between the solid and pyridine was measured.

The calorimeter cell is a 100 ml, silvered, double-walled vacuum flask equipped with a

heater coil and thermistor. The thermistor is calibrated prior to each experiment and is

connected through a digital / analog converter to a computer which calculates the heat of


Adsorption Studies

Adsorption of pyridine from cyclohexane was combined with calorimetric data to

yield a combined Cal-Ad4 analysis. Such an analysis provides information about the

quantity, equilibrium constant for binding, and the enthalpy of interaction for the various

types of sites on the solid. The adsorption experiment is similar to the calorimetric

analysis in that the same solvent to solid ratio is used. UV-VIS measurements were made

on a Perkin-Elmer Lambda 6 UV-Vis spectrophotometer at a wavelength of 252 nm with

0.1 cm quartz cuvettes. The total number of moles of probe adsorbed per gram of solid

adsorbent, n, was calculated using Equation 4-1.

=(C C,)V(0.OOL mL) Equation 4-1
n = Equation 4-1

where Ci is the initial probe concentration (mol L '), C,q is the equilibrium probe

concentration (mol L1), V is the volume of solution added (mL), and m is the mass of

adsorbent (g). Total adsorption isotherms were obtained by plotting n vs. C.,.

Thermogravimetric, InfraRed, and Carbon, Hydrogen, Nitrogen Analyses

Thermogravimetric analyses were conducted on a DuPont 2000 TGA. Typical

heating rates were 10C per minute ranging from ambient to 10000C. IR studies were

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EZFQFD5L1_EQK4H0 INGEST_TIME 2013-02-07T19:29:31Z PACKAGE AA00012974_00001