|Table of Contents|
Table of Contents
List of Tables
List of Figures
Chapter 1. An NMR method for probing the micro-, meso-, and macropore behavior of porous solids
Chapter 2. Extension of the unified solvation model to acceptor solvents and the interpretation of solvent controlled reactions
Chapter 3. Conclusions
Appendix. Additional data
THE USE OF CHEMICAL PROBE MOLECULES
TO INVESTIGATE THE NATURE OF
ADSORBENTS AND SOLVENTS
DONALD CHRISTOPHER FERRIS
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
God gave us two ears and only one mouth so that we may listen twice
as often as we speak.
He also gave us two feet so that if we choose to place one of them
in our mouth we may still make a graceful exit with the second.
Donald "Bueller" Ferris
This work is dedicated to my family, whose love and faith in me were
stronger than my own, without which this work would never have come to
pass. This work is also dedicated in memory of my mother, Adele Carla
Ranieri Ferris (19 June 1937 to 18 November 1978), and my sister,
Catherine Ferris (10 December 1959 to 1960).
It is often said that a journey begins with one step. The road to
a dissertation is both a long and difficult journey. Many have undertaken
this journey before, and certainly, many will travel after, but this
particular journey was my own, though it was hardly traveled alone. The
path was far from straight and many helped guide me when I strayed from my
course. It is impossible to mention everyone who assisted me. For all who
have helped me, I sincerely thank them from the bottom of my soul. Even
though the actual stops an the journey were my own, the decisions and
choices reflect that I am an amalgam of the influences on my life due to
My journey began somewhere in middle school and high school where I
had several good educators. I would especially like to thank Mr. Ira
Simpson, my sixth grade teacher. He especially encouraged me to reach
beyond my potential. Most certainly my science teachers, Mr. McGarrity,
Mr. DeLorentis, Mr. Chroman, and Mr Ames, all of whom were well qualified,
were informative and interesting. I would like to thank Mrs. Boss for
looking past the troubled handwriting to the person on the other side. It
was probably Mr. Dixon's fever and enthusiasm during A.P. History that has
kept my interest in current events going. Finally, I would like to thank
Mr. White, who taught me to look at other cultures and people without
The destination of my journey came into focus at Ithaca College. I
would like to thank the entire faculty and staff of the Chemistry
Department, who taught me that asking "Why?" can lead to a very
interesting life. In particular I would like to thank Dr. Anatol Eberhard
and Dr. Bill Bergmark, who allowed me to do research with them; Drs. Heinz
and Judy Koch for information, insight and homemade ice cream; Dr. Glenn
Vogel for harassment and an occasional good word; and certainly Jim
MacNiel, who taught me how to use many an instrument and then fixed them
after I broke them.
Indeed the most challenging and difficult portion of the journey
began when I arrived at UF. There is a wealth of knowledge in the faculty
and staff in the Department of Chemistry. It has been quite a privilege
to have been in contact with several of the people here. I came
particularly to work for a specific person, who has been mentor, guide,
disciplinarian, friend, advisor, parent, teammate, opponent, source and
target of harassment, and, by the way, chemist. I have matured not only
as a chemist but as a human being. I gained more than an insight into
chemistry from Dr. Russell S. Drago. I just hope I do not succumb to
using that silly underhanded shot in basketball when I turn "49."
On a personal note there are a few people I would like to mention.
During my time at UF I have come to learn what a special treat it is to
know Ruth Drago. I do not think I (or any member of the "extended Drago
family," for that matter) will ever be able to capture in words what she
has come to mean to me. I will always cherish her smile, words of
encouragement, motherly advice and dynamic personality. There are several
past and present members of the Drago Group who I want specifically to
mention, although they all have been great. In particular they are my
roomiess" at the "Thunder Dome," Mark "Citizen" Barnes, Gerry "I can
figure out a nickname for anybody" Grunewald and Steve "Wally" Showalter;
my fellow I.C. alumni and "E and C'ers" Ngai Wong and Andy Dadmun; my
"classmates" Steve "Beav" Petrocious and Mike "The Round Mound" Naughton;
my labmates Chris "Air" Chronister, Steve Jorge, and Melissa "Bunny"
Hirsch; the secretarial staff of Maribel Lisk, April Kirck, and Diana
Williamson; a slew of postdocs including Doug "I really should go home
now" Burns, Dave "What Visa" Singh, Phil "Why you say that" Kaufman and
Bob "Out there somewhere" Beer; my Army buddies Brigadier General (Ret.)
Jim Ramsden and Captain Karen Frank; and my workout and thesis writing
partner Johnny "Cyanide" Hage. Several people I met during my stay in
Gainesville have been quite enjoyable, mainly my buddies from Players:
Kim, Pat, Val, Terry, Liz, Kevin and Gil. I would also like to thank the
past and present members of the 410th Quartermaster Detachment, who were
a pleasure to have served with for four years and to have commanded for 41
As my dedication indicates, this journey would have never even come
close to being completed without the support of my family. My father, J.
David Ferris, has always supported me no matter what I have chosen to do.
I am especially proud of the close relationship I share with my brothers
David and Jamie, and my sister, Kim. My love of science came from my
mother, who was attending college to become a science teacher when she
opted for a career as a mother instead. The rest of my extended family
has been wonderfully supportive. I can never thank them enough.
I would like to thank someone who has really made the last year
enjoyable. I find with every day that my dependence on her being with me
increases. Elizabeth Grzybinski has become a source of strength and
inspiration to me, and I look forward to being with her for a very long
Finally, I have become a true believer in "pet therapy." Our two
dogs, Sam and R.J., two cats, Ginny and Calvin, and fish, Ginsu, have
provided relaxation, fun and unconditional love. I have found that no
matter what type of day I have had or how impossible something seems,
whenever I go home there is "someone" happy to see me.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS . . . . . . . . . . . . .
TABLE OF CONTENTS . . . . . . . . . . . .
LIST OF TABLES . . . . . . . . . . . . .
LIST OF FIGURES . . . . . . . . . . . . .
ABSTRACT . . . . . . . . . . . . . . .
1 AN NMR METHOD FOR PROBING THE MICRO-, MESO-, AND MACROPORE
BEHAVIOR OF POROUS SOLIDS . . . . . . . . .
Background .. . . . . . . . . .
Carbonaceous Adsorbents . . . . . . . .
Physical Adsorption Principles . . . . . .
Adsorption Time Dependence . . . . . . .
Characterization of Carbonaceous Materials . . .
The Use of Nuclear Magnetic Resonance (NMR) . . .
Experimental . . . . . . . . . . . .
Solutions.. . . . . . . . . . ..
Solid Materials .. .......................
Modification of Materials......................
Instrumentation . . . . . . . . . .
NMR Studies....... .. .. . . . . . . . ..
The general NMR experiment . . . . . .
Concentration studies . . . . . . .
Time . . . . . . . . . . .
Temperature . . . . . . . . . .
T, . . . . . . . . . . . .
Vapor study . . . . . . . . . .
Curve resolution . . ..........
Blanks . . . . . . . . . .
Results and Discussion . . . . . . . .
Inorganic Materials ..... . . . . . ..
Silica gel . . . . . . . . . .
Kieselguhr . . . . . . . . . .
Zeolites . . . . . . . . . .
Carbon-based Adsorbents . . . . . . . .
Ambersorb* 564 . . . . . . . . .
Ambersorb* 572 . . . . . . . . .
Ambersorb 348f .. . . . .... .. .....
MTV1102 ........ ....................
Ambersorb 563 .... ...........
Activated carbon (Baker Food Grade, powdered) .
Summary of the qualitative CH3CN studies . .
Modified Ambersorb materials.. . . . ..
Quantification of the Method . . . . . . .
The process . . . . . . . . . .
CH3CN with 572 . . . . . . . .. 56
C6H6 with 572 . . . . . . . . .. 58
Chemical Warfare Agent Simulants . . . . . 60
DMMP . . . . . . . . . . .. 61
TMP . . . . . . . . . . .. 65
2 EXTENSION OF THE UNIFIED SOLVATION MODEL TO ACCEPTOR SOLVENTS
AND THE INTERPRETATION OF SOLVENT CONTROLLED REACTIONS . 70
Background . . . . . . . . . . . . 70
Previous Models.... . . . . . . . . 70
The Universal Solvation Model . . . . . .. 75
Results and Discussion. . . . . . ......... 78
Systems Involving Non-Specific Interactions.. ... . 78
An Experimental Procedure for Factoring Specific and
Nonspecific Contributions to AX . . . . .. 85
Data Fit for Acceptor Solvents . . . . . .. 92
The Cavity Term in Solvation Effects . . . .. 103
A One-Donor Parameter Fit of the Specific Interaction 103
Adding to the USM . ..... ................ 104
Adding new acceptor solvents .......... 104
Adding new probes . . . . . . . 105
Using the Parameters to Interpret Solvent Influenced
Reactivity . . . . . . . . .. 110
Enantiomeric excess . . . . . . . 110
Analysis of solvent effects induced in
physiochemical properties of coordination
compounds . . . . . . . . .. 112
Fe(phen)2(CN)2 . . . . . . . .. 113
Cu tmen acac . . . . . . . . 122
Ni tfd phen... . . . . . . . 122
Decarboxylation of benzisoxazole-3-carboxylate
ions . . . . . . . . . .. 125
Electron Transfer Processes . . . . . .. 149
Transition state theory. .... ....... 150
Solvent friction correction to TS theory . 153
The two-dimensional Sumi-Marcus approach . 156
Commonalities of the two models .. . . . 159
The unified solvation model and electron transfer
processes . . . . . . . . .. 160
Cp2Co(+" Self-Exchange . . . . . .. 162
Ultrafast Spectroscopy . . . . . .. 176
Experimental . . . . . . . . . . .. 187
Purification of Reagents. . .......... . 187
Betaine Shifts with Solvent Composition . . . 190
Burgess Dye Shifts with Solvent . . . . .. 190
Calculations . . . . . . . . . . .. 190
3 CONCLUSIONS . . . . . . . . . . . .. 195
APPENDIX ADDITIONAL DATA . . . . . . . . . .. 196
REFERENCES . . . . . . . . . . . . .. 212
BIOGRAPHICAL SKETCH . . . . . . . . . . .. 220
LIST OF TABLES
1-1 Physical Properties of Some Carbonaceous Material . . . 7
2-1 S' Parameters for Solvents . . . . . . . .. 80
2-2 P and W Parameters for Probes . . . . . . . .. 81
2-3 Specific and Non-Specific Contributions to Am from the Data in
Figure 2-3, Figure 2-4, and Figure 2-5 . . . . .. 90
2-4 Data Fit for Acceptor Solvents . . . . . . .. 94
2-5 Parameters for Estimating Specific and NonSpecific Solvating
Properties of Acceptor Solvents . . . . . . ... 99
2-6 Probe Acceptor Parameters . . . . . . . .. 100
2-7 Fit of Reported Data for Benzylalcohol to Equation (2-7) 106
2-8 Fit of Literature Data for Reported Probes to Equation (2-
7) . . . . . . . . . . . . . .. 107
2-9 Solvation Model Estimation of Enantiomeric Excess. .... ill
2-10 Solvent Effects on the Electronic Transitions of
Fe(phen)2(CN)2 . . . . . . . . . . .. 114
2-11 Solvent Effects on the Electronic Transitions of
Fe(phen)2(CN)2 . . . . . . . . . . .. 116
2-12 Electronic Transition of Cu tmen acac2* in Various Solvents
(kK) . . . . . . . . . . . . .. 121
2-13 Solvent Shifts of Ni(tfd)(phen) . . . . . . .. 123
2-14 Rate Constants for 6-Nitrobenzisoxazole-3-carboxylate Ions In
Several Solvents . . . . . . . . . . 128
2-15 Solvents From the Dissociated Ion Pair Category Fit to
Equation (2-13) . . . . . . . . . .. 135
2-16 Solvents From the Ion Pair Category Fit to Equation (2-14) 136
2-17 Fit of Specific Solvation with Protic Solvents . . .. 143
2-18 Solvent Dependence of the Coboltacene/Coboltacenium Self
Exchange Reaction . . . . . . . . . .. 166
2-19 Solvent Dependencies and USM Calculations for Cp2Fe(+^' Self
Exchange Reactions . . . . . . . . . .. 173
2-20 Solvent Dependencies and USM Calculations for L2H('M Self
Exchange Reactions Rates . . . . . . . . 174
2-21 Solvent Dependence of the Photodynamics of S, Bianthryl . 182
2-22 Solvent Dependent Oscillator Strengths and Free Energy
Parameters Obtained from the Static Absorption and Emission
Spectra of ADMA . . . . . . . . . . . 183
2-23 Electron Transfer Rates of Betaine . . . . . .... 188
A-1 Data Fit for Parameters in Table 2-1 and 2-2 . . .. 197
LIST OF FIGURES
1-1 Conceptual diagram of the pore structure of carbonaceous
adsorbents . . . . . . . . . . . . 3
1-2 Proposed chemical intermediates for the pyrolysis of
sulfonated styrene/divinyl benzene .. . . . . . . 5
1-3 Qualitative representation of the van der Waals force between
an adsorbate molecule and a planar surface as a function of
separation in adsorbate molecular diameters. Positive Y is
repulsion, negative Y is attraction. . . . . . . 6
1-4 Illustration of the effect of two surfaces as in Figure 1-3;
2, 3 and 4 molecular diameters apart. Positive Y is repulsive,
negative Y is attractive . . . . . . . . .... 10
1-5 Adsorbent site competition between solute and solvent. A,
adsorption potential for solvent; B, interaction potential
between solute and solvent; C, adsorption potential for
solute . . . . . . . . . . . . . .. 12
1-6 The adsorption pathway of an adsorbate from the bulk to the
interior of a carbonaceous adsorbent (not drawn to scale). 15
1-7 Adsorbate distribution within a adsorbent for A, transport
pore limited diffusion; B, micropore limited diffusion; C,
the intermediate case of similar transport pore and micropore
diffusion rates . . . . . . . . . . .. 18
1-8 Relative location of sample in the NMR window. . . . 28
1-9 1H NMR spectrum of a 0.3 M CH3CN solution in CC4 with dried
silica . . . . . . . . . . . . . .. 31
1-10 1H NMR spectrum of a 0.3 M CH3CN solution in CCI4 with
Kieselguhr . . . . . . . . . . . . .. 33
1-11 1H NMR spectra of a 1.0 M CH3CN solution in Cc4 with NaY. 34
1-12 1H NMR spectrum of a 0.3 M CH3CN solution in CCl4 with NH4Y. 35
1-13 'H spectrum of a 0.3 M CH3CN solution in CC14 with HY ..... 36
1-14 1H spectra for 1.0 and 2.0 M CH2Cl2 in CC4 with 564. ... 38
1-15 1H NMR spectrum of a 1.0 M CH2Cl2 solution in CCl4 on crushed
564. . . . . . . . . . . . . . . 41
1-16 'H NMR spectra of 2.0 M CH3CN solution in CCl4 with 564. The
bottom is a "fresh" sample and the top is a 100 hour old
sample . . . . . . . . . . . . ... . 42
1-17 1H NMR spectra of solutions of 1.0 M CH3CN, 1.0 M CH2Cl2 and 0.9
M in each in CCl4 with 564 . . . . . . . ... 44
1-18 1H NMR spectra for 0.05, 0.2, and 0.5 M CH3CN solutions in CC4
with 572. The bottom spectra is the 0.05 M solution, the top
is the 0.5 M solution . . . . . . . . ... 45
1-19 1H NMR spectrum of a 0.5 M CH3CN solution in CC4 with 572. 47
1-20 1H NMR spectrum of a 0.5 M CH3CN solution in CC4 with 348f. 49
1-21 1H NMR Spectrum of a 0.5 M CH3CN solution in CC4 with MTV.
1172 . . . . . . ........................... . 50
1-22 IH NMR spectrum of a 0.5 H CH3CN solution in CCl4 with AC. . 52
1-23 Acetonitrile adsorption based on the "free" acetonitrile peak
area in the 0.5 M solutions . . . ............... . 53
1-24 'H NMR spectra for a 1.0 M C^ solution in CCl4 with 564 and
564-Cl . . . . o. . ......................... . 55
1-25 The amount of CH3CN adsorbed from solutions of various
concentrations into the different pores as determined by NMR
and GC for samples of various ages . . .............. 57
1-26 The amount of CA6 adsorbed in the various pores as determined
by NMR and HPLC for a one day sample of various solutions. 59
1-27 The 31p NMR spectrum of a 1.0 M TEP solution in CC4 with
572. . . . . . ............................. 62
1-28 The 31P NMR spectrum of a 0.3 M DMMP solution in CC4 with
572. . . . . . . . . . . . . . . 63
1-29 The 1H NMR spectra of 0.3 M DMMP solution in CC14 with (from
bottom to top) 572, 348f, AC, and MTV 1102 . . . .... 64
1-30 The IH NMR spectra of several solutions of TMP in CCl4 with
572. . . . . . .............................. 66
1-31 Results of quantitative NMR studies of TMP solutions with 572
(top) and 564 (bottom) . . . . ................... 67
2-1 Pyridinium N-phenoxide Betaine Dye . . . . o.... . 72
2-2 Stilbazolium Betaine and its Derivatives ........ 73
2-3 The Shift in the Electronic Transition of Betaine with
Solvent Composition in Methanol, o-Dichlorobenzene ..... 87
2-4 The Shift in the Electronic Transition of Betaine with
Solvent Composition for Alcohol, o-Dichlorobenzene Mixtures.
Key to solvents: t-butanol, A; 1 butanol, V; 1 octanol, 0;
ethanol, 0. .......... ........................ 89
2-5 The Shift in the Electronic Transition of Betaine with Solvent
Composition for Methylene Chloride, o-Dichlorobenzene
Mixtures . . . . . . . . . . . . ... . 91
2-6 Calc. vs. exp. values of Burgess Dye using data from ref 80.
The line represents the ideal case . . . . . .... 119
2-7 Decarboxylation of benzisoxazole-3-carboxylate ion . . 126
2-8 Plot of the natural log of the decarboxylation rates vs. S.'
Closed squares are for aprotic solvents, open squares are for
protic solvents. Label numbers match table ......... 130
2-9 Plot of the "right" side of Figure 2-8. The line is
calculated using ln(k) = S'P + W . . . . . .... 132
2-10 Ln(k) vs S' for the "left" side of Figure 2-8. The line is
calculated from ln(k) = S'P + W . . . . . .... 134
2-11 Plot of ln(k) vs S' with the lines from Figure 2-10 and
Figure 2-9 added. Solvents from Figure 2-10 have not been
corrected for specific interaction (indicated by the solvents
deviation from the line) . . . . . . . .... 139
2-12 Plot of calculated ln(k) vs. the experimental ln(k). The line
is the ideal case. The A and v symbols represents the
calculated value for solvent 9 and 11 using equations (2-13)
and (2-14) respectively. . . . . . . . . 142
2-13 Free energy surfaces as functions of solvent coordinate q. 151
2-14 Intersection of product and reactant FESs (Figure 2-13). 152
2-15 Contour plot of the two-dimensional FES as a function of a
diffusive solvent coordinate X and a low-frequency
intramolecular coordinate Y . . . . . . .... 157
2-16 Plot of the function (1/n2-1/e,) vs. S'. . . . . .. 161
2-17 Two most likely symmetries for the transition state for the
self exchange of CpM(+/O) . . . . . . . .... 164
2-18 Schematic orbital diagram for Cp2M. Orbital symmetry based on
pseudo Dh. Level ordering based on INDO calculations.
Electron filling based on M = Fe . . . . . .... 165
2-19 Plot of the In(k61) vs S' for the solvents. The line is the
fit of the donor solvents (sans 2,4 and 8) to only ln(kj) =
1.19 S' + 13.83. . . . . . . . . .. . 168
2-20 Equilibrium between the bianthryl isomers LE and CT. . 177
2-21 Plot of vC vs. S' for the solvent. Filled squares represent
aprotic solvents. The open squares represent protic solvents.
Numbers correspond to table 2-20 . . . . . .... 179
2-22 Plot of ln(kcT) (left hand Y axis and filled squares) and
In(Kq) (right hand Y axis and filled circles) vs. S' of the
solvents. Numbers correspond to table 2-20. . . . 180
2-23 Plot of f vs. S'. Numbers match those in table 2-21. . 181
2-24 plot of k (solid squares) and AG (solid circles) vs. S" for
the solvent. Numbers correspond to table 2-21 ....... 184
2-25 Schematic diagram illustrating the relative positions of the
FESs for LE, CT and the ground state in the "inverted
region". . . . . . . . . . . . .. 186
2-26 Plot of k, vs. S' for the solvent. Numbers correspond to
those in table 2-22 . . . . . . . . .... 189
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
THE USE OF CHEMICAL PROBE MOLECULES
TO INVESTIGATE THE NATURE OF
ADSORBENTS AND SOLVENTS
Donald Christopher Ferris
Chairperson: Russell S. Drago
Major Department: Chemistry
Despite the wide use of carbon-based adsorbents for many purposes,
the nature of their interior still remains a "black box." Though
porosimetry provides useful information, there is still a desire to know
more. In many reactions, results that are unexpected are sometimes waved
away as "solvent effects." These effects can cause reaction rates to vary
by several orders of magnitude.
In the first portion of this dissertation an indirect method for
analyzing the pore dimensions and distribution in solid materials is
presented. Small, simple compounds are allowed to diffuse into the porous
network of carbon-based materials. The nuclear magnetic resonance (NMR)
spectrum of the "probe" molecules is obtained. The observed change in the
spectrum of the probe in the presence of the solid material from the
spectrum of the probe in the absence of the solid material is a measure of
the effect the pore structure has on the probe. The change in the NMR
spectrum of the probe occurs only for organic materials. A comparison
with more well-characterized inorganic materials also is done to
illustrate and compare the method. The results can be quantified by
carefully measuring the amount of probe adsorbed by the solid. In
addition, the method is used to analyze the adsorption of chemical warfare
agent simulants by some adsorbents.
The second portion of this work looks at how the UV-visible, NMR,
IR, enantiameric excess, etc. of probe molecules change in different
solvent mediums. Both solid and liquid probes are dissolved into the
solvent. The nature of the solvent can affect spectral shifts, reaction
rates, and mechanism pathways. A model is discussed that parameterizes
these changes. The highly interacting hydrogen bonding solvents are added
to the model, which then is used to analyze several organic reactions and
electron transfer reactions to interpret the nature of the solvents'
pronounced effect on these phenomena.
AN NMR METHOD FOR PROBING THE MICRO-, MESO-,
AND MACROPORE BEHAVIOR OF POROUS SOLIDS
Granular activated carbons (GACs) and carbonaceous adsorbents have
been used in the purification of groundwater for some time.1 These
materials have interesting applications besides that of an adsorbent.
Recent work has demonstrated that carbonaceous adsorbents are novel
catalysts and catalyst supports.2'34 Carbonaceous adsorbents and GACs can
be prepared from several sources. GACs are prepared from naturally
occurring carbonaceous materials such as petroleum coke,5 anthracite coal,
wood char, coconut husks, compacted peat moss, etc.6 Carbonaceous
adsorbents use synthetic polymeric solids as their precursors (often
referred to as polymeric adsorbents).6 The final product is created by
pyrolysis, which is the process of heating the material in an inert
atmosphere. Pyrolysis begins around 300 C and is occasionally not
completed until 2700 C.1 The use of synthetic functionalized
macroreticular (highly cross-linked) polymer resin beads allows for
greater control of the carbon skeleton in the pyrolized product.7"
Higher surface areas can be obtained from practically any
carbonaceous solid through an activation process. Activation creates
small voids (pores) in the material by the oxidation and subsequent
removal of some of the solid.' Typical activation reactions involve the
use of steam, air, and/or carbon dioxide 'at 700 to 900 C.1 In the case
of the carbonaceous adsorbents, the nature of the cross-linking in the
synthetic polymer, as well as the polymer itself, can have a dramatic
effect on the final pyrolyzed product. The use of polysulfonated
styrenes/divinyl benzenes(dvb) has proven to yield adsorbents with
remarkably reproducible porosity and surface area.1"9 Research has also
shown that the choice of pyrolysis temperature and technique can alter the
pore size distribution and change the molecular sieving property of the
An adsorbent (organic or inorganic) is a solid that is characterized
by high surface areas and is capable of concentrating molecules (called
adsorbate molecules or adsorbates) onto the solid. Surface areas are
expressed in units of square meters per gram of dry adsorbent. Useful
adsorbents have surface areas from ca. 400 to 2000 m2/g (the area of a
football field is 4180 m2).1 The van der Waals force between the solid
and the adsorbate is one of the driving forces that concentrates the
adsorbate on the solid and removes it from the liquid or gas in
equilibrium with the solid.
The size and shape of the pores in carbonaceous adsorbents have
impact not only on the adsorption capability of the solid but also in its
molecular sieving ability and speed of absorption.1 Unlike zeolites, which
have cylindrical pore openings, carbonaceous adsorbents have slit-shaped
pores.1 The slit shape is demonstrated by the selective adsorption of
benzene (3.7 A thick x 7.0 A wide) and the exclusion of carbon
tetrachloride (roughly 6.0 A diameter sphere)1 and is shown in Figure 1-1.
Whereas zeolites tend to prefer linear molecules, carbon molecular sieves
incline to have the order planar > linear > branched.
Macroreticular polymer precursors have latent porosity due to the
cross-linked polymer. Indeed many precursors are ion exchange resins,
which by their very nature are made to be quite porous for effective ion-
exchange with the polymer functionality.' When the polymers are pyrolyzed
two things can happen to the carbon material. The first is that the
material "melts" (fuses), which slowly graphitizes the polymer with the
loss of the original porosity. The second is that the material "chars"
0 Carbon Tet
Figure 1-1 Conceptual diagram of the pore structure of carbonaceous
and keeps the majority of its original porosity.1 Examples can be found
of both extremes. For the majority of carbonaceous adsorbents both
processes occurred during the pyrolysis of the polymer precursor. The
importance of the porosity in the precursor material is that these pores
can act as conduits during the pyrolysis process to allow an escape route
for volatile organic. If the porosity is maintained in the final solid,
they can also act as conduits for adsorbate molecules. The evolution of
volatile organic leads to the creation of small pores in the final
The sulfonated divinyl benzene precursors of the Rohm & Haas Co.
(vida supra) are believed to follow the scheme shown in Figure 1-2.
During the pyrolysis the majority of sulfur is lost as SO. Hydrogen
content decreases with increasing temperature. The resulting materials
(trade name Ambersorbe) are 90+% carbon, 1-2% hydrogen and the remainder
mainly oxygen with some residual sulfur.
Table 1-1 shows several carbonaceous materials and some relevant
physical properties. The different pore sizes are based on the IUPAC
definition, which states (when N2 is the adsorbate used for pore volume
measurements) that micropores are less than 20 A, mesopores are between 20
and 500 A, and macropores are larger than 500 A.11 Note in table 1-1 that
the GAC and AX21 (a material based on the pyrolysis of petroleum coke)
have very little macroporosity and are almost entirely microporous. This
is in contrast to the various Ambersorbe materials, which contain a
distribution of pores. It should be noted at this time that the Ambersorbe
materials do not contain a continuous distribution of pores. They have
micropore and small mesopores (< -60 A), then a range of pore sizes where
no actual pores exist, then large mesopores and macropores.
Physical Adsorption Principles
The force responsible for the phenomenon of adsorption at
interfacial surfaces, as well as deviations from ideal gas behavior,
Figure 1-2 Proposed chemical intermediates for the pyrolysis of
sulfonated styrene/divinyl benzene.
Figure 1-3 Qualitative representation of the van der Waals force between
an adsorbate molecule and a planar surface as a function of
separation in adsorbate molecular diameters. Positive Y is
repulsion, negative Y is attraction.
Physical Properties of Some Carbonaceous Materials
Property 563a 564" 572a KUREHAb AX21c Act Carbd
BET 550 550 1100 1000 2000+ 1100
Micro 0.23 0.24 0.41 0.53 1.7 0.58
Meso 0.14 0.13 0.19 0.02 0.700 0.07
Macro 0.23 0.21 0.24 0.02 0.01
Hydro- high high low high
Ash % 0.08 0.22 1.84 5.56
a) These materials are Ambersorbe materials of the Rohm & Haas Co.
Amount of activation done to the material 563 < 564 < 572. Data
provided by Rohm & Haas Co. Bead material
b) Product of Kureha Chemical Industry Company, Ltd. Data provided by
Rohm & Haas Co. Bead material
c) Product of the Anderson Development Company. Data provided by Rohm
& Haas Co. Powder material
d) GAC produced by Calgon. Data provided by Rohm & Haas Co. Powder
e) For this material it is difficult to distinguish the break up of
porosity between the macro and mesopores. The value given is the
total for both types. It is believed that it is not an even
distribution, but that more volume is in the mesopores.
condensation of gases to liquids, and the crystalization of liquids to
solids is the van der Waals force. (Sometimes it is referred to as London
dispersion forces because they are also responsible for the phenomenon of
optical dispersion.) The van der Waals force arises from the rapidly
fluctuating electron density in one molecule inducing a complementary
electrical moment in a near neighbor, which results in an attraction
between the two molecules. The interactions are weak and fall off as r4
dependence with distance.12 However, within 2-3 molecular dimensions the
attractive force can be significant (2-15 kcal/mole).' This is shown
qualitatively in Figure 1-3 for a molecule interacting with a planar
surface. The strength of the interaction is also directly related to the
polarizability of both the support and the adsorbate. Since
polarizability is a function of r3 and the area of contact is directly
related to the strength of the induced dipole, the size or footprint of
the adsorbate molecule is also important.13 Once the adsorption site is
filled with an adsorbate, the van der Waals interaction of a second
adsorbate with the solid is reduced dramatically by the distance (vida
supra). The adsorption of the second layer is more closely related to the
latent heat of vaporization, which is related to the van der Waals
interaction between two adsorbate molecules. When the adsorption
potential of the solid with the adsorbate is greater than the latent heat
of vaporization, all the available surface will be covered by a monolayer
of adsorbate molecules. This phenomenon is used frequently to measure
surface areas (table 1-1)."
Electrostatic forces can have significant impact on adsorption when
polar molecules are adsorbed. The dipole at the surface is now static
rather than oscillating and is similar to that of a capacitor or an
electrode. For many inorganic oxides, salts, and zeolites the
electrostatic force can be the dominating force for attraction of polar
adsorbates. This interaction is less sensitive to the distance from the
surface (r3 vs r6).12 Dipole moments can be induced in nonpolar
adsorbates. Though the induced dipole is not large for nonconducting
solids with nonpolar molecules, the effect can be quite large for a
conducting material like graphite with nonpolar adsorbates.14
A recent article by Steele reviews the state of molecular mechanics
and modeling of several surfaces to include graphite and amorphous
carbon.15 For graphite the surface (using Kr as the model adsorbate)
calculations yield that the surface-averaged adsorption energy is -3.83
kcal/mol and has a relatively flat periodic adsorption (the strongest
sites correspond to the space over the carbon hexagons).15 The amorphous
carbon surface has adsorption sites ranging in energy from -1.25 to -3.32
kcal/mol (corresponding to the highest and lowest points on the surface,
which is varied 1.83 A).15
Effective adsorbents are those with a large volume of very small
pores. A qualitative estimation of an average pore size can be obtained
if one assumes that the pores are open-ended cylinders. Though this is
not an accurate assumption (recall that for carbonaceous materials the
pores are slit-like, vida supra), average pore diameters (d) can be
related to the total pore volume (V) and surface area (S) by equation 1-1.
Average pore sizes calculated by equation (1-1)
d = 4V/S (1-1)
can be notably erroneous because pore volume tends to exist predominantly
in larger pores while the surface area is mainly in smaller pores. The
coarse calculation, however, does show that adsorbents with high surface
areas will have pore sizes on the order of molecular dimensions.'
The range of pore sizes found in high surface area adsorbents
corresponds to distances where electrostatic and van der Waals forces are
effective. The presence of a second surface can greatly alter the
adsorption potential of a molecule on a surface as shown by Figure 1-4.1
The adsorption of molecules occurs predominantly in small pores for two
reasons. The first is that the energetic are favorable as illustrated in
Figure 1-4. The second is that most of the surface area is contained in
Figure 1-4 Illustration of the effect of two surfaces as in Figure 1-3;
2, 3 and 4 molecular diameters apart. Positive Y is repulsive,
negative Y is attractive.
small pores. This can be illustrated with an example from reference 1.
If one has an adsorbent with a surface area of 1000 m2/g and a pore volume
of 0.7 cm2/g then using equation (1-1) an average pore size is calculated
to be 28 A. One can also back calculate using equation (1-1). That is,
if the average pore size and pore volume are known, then the surface area
can be estimated. If we assume that there are pores of two sizes, 14 and
1000 A, each with equivalent pore volumes, 0.35 cm2/g, the calculated
surface areas arising from these pores are 1000 m2/g and 14 m2/g,
respectively. This simple calculation shows that 98.6% of the surface
area is contained in the small pores, though they correspond to only have
half the pore volume.
As with any chemical phenomenon, equilibrium is reached in equation
(1-2) when the rate of the forward reaction,
A + S = A-S + heat (1-2)
adsorption, is equal to the rate of the reverse reaction, desorption
(where A is the adsorbate and S is the surface). The average residence
times for molecules adsorbed out of the vapor phase has been calculated as
a function of the energy of adsorption.16 The residence times range from
10"13 to 10.2 seconds for typical range of physical adsorption energies (vida
supra). Therefore, a rapid exchange is going on between the adsorbed and
free states, which implies that for readily available surfaces equilibrium
is obtained quickly. Additionally, there can be rapid diffusion along the
solid surface by the adsorbed species. Indeed a two-dimensional van der
Waals gas law model for the adsorbed species is the most successful for
describing physical adsorption.' The implication is that adsorbed species
experience unlimited mobility along the solid surface. In order to
localize an adsorbed film of a gaseous adsorbate, temperatures below that
of liquid N2 are required.17
The adsorption of solutes from solution is profoundly influenced by
the forces depicted in Figure 1-5.1 The ability of a particular solute to
adsorb is based on the difference in the adsorption potential between the
Figure 1-5 Adsorbent site competition between solute and solvent. A,
adsorption potential for solvent; B, interaction potential
between solute and solvent; C, adsorption potential for
solute and the solvent (forces C and A in Figure 1-5, respectively). The
ability of an adsorbent to adsorb solutes is limited in the case where the
solid has a high affinity for the solvent (A > C). An example is the low
adsorption ability of zeolites (a polar adsorbent) for solutes in water (a
polar solvent). Conversely, GACs and carbonaceous adsorbents have high
adsorption capacities in water due to the low adsorption potential energy
required by the solute to displace a water molecule (C > A).1
Merely having a high affinity for the solute is not sufficient by
itself to insure a high adsorption capacity. The potential for adsorption
of the solute for the surface must also be greater than the attractive
force between the solute and the solvent. There exist an inverse
relationship between the adsorbate solubility and the adsorption capacity.1
Essentially, less soluble materials will be more readily adsorbed than
more soluble materials.
Adsorption from solution is limited to a monolayer because solute
interaction with the solvent is usually greater than solute-solute
interaction if the solute is soluble. By analogy to the vapor phase (vida
supra), the attraction of solutes in the first monolayer for unabsorbed
solute molecules can be assumed to be equal to the attraction of a surface
of a pure liquid solute for a dissolved solute molecule.1 However, the
pure soluble liquid solute will spontaneously dissolve at any
concentration below the saturation concentration. Therefore, adsorption
from solution beyond the first monolayer occurs only rarely.'6
Until now we have strictly dealt with "adsorption" and not
"absorption." Absorption occurs when an adsorbate can penetrate into the
solid rather than remain constrained to the solid surface and is often
difficult to distinguish experimentally from adsorption.1 If the
possibility exists for both phenomena to occur then the more general terms
of sorption, sorbate and sorbent are sometimes used. Absorbance is most
readily detected by the occurrence of a change in the apparent volume of
the adsorbent. Activated carbon swells very little in pure liquid
sorbates indicating that absorption is not an important contribution to
the sorption capacity. However, polymeric and carbonaceous adsorbents can
swell significantly (50% in some cases). Therefore, absorption has been
suggested as being important in the overall sorption capacity of these
adsorbents.1 For the remainder of this document the terms adsorption,
adsorbate and adsorbent will be used and where appropriate a reminder
about absorption will be made.
Temperature, molecular size, and concentration of adsorbate can also
influence the adsorption equilibrium. It is readily seen in equation (1-
2) that high temperatures will favor desorption. As mentioned previously
the size of the molecule is important. In general larger molecules will
be adsorbed more strongly than smaller ones. This trend can be reversed
if pores exist that inhibit the diffusion of the molecule (vida supra and
vida infra). The concentration of nonadsorbed molecules in the solution
or gas phase in contact with the solid is the most important variable
influencing equilibrium adsorption capacity.' A typical adsorption
isotherm from solution shows a rapid increase in the amount adsorbed with
increasing concentration at low concentrations. At higher concentration
the amount adsorbed as more solute is added usually levels off.
Adsorption Time Dependence
For surfaces that are readily available equilibrium is established
rapidly (vida supra). However, for granular adsorbents (like the Rohm and
Haas materials) even a well-stirred solution can take days or weeks to
reach equilibrium.' This is due to the slow transfer of the adsorbate from
the bulk solution outside the particle to the active adsorption sites
inside the particle. Nearly all the adsorption sites are inside the
material with the number of sites on the exterior surface being
Prior to adsorption, the adsorbate is dispersed throughout the bulk
solution. The rate of adsorption, therefore, is determined by the rate of
Figure 1-6 The adsorption pathway of an adsorbate from the bulk to the
interior of a carbonaceous adsorbent (not drawn to scale).
transfer of the adsorbate from the bulk solution to an active adsorption
site within the particle. Figure 1-6 qualitatively shows the steps an
adsorbate must follow in order to be adsorbed.' First, the adsorbate must
cross from the bulk through a stationary film surrounding an adsorbent.
Second, the adsorbate must pass through the transport pore (macropore and
large mesopore) region. Next it must pass through the micropore (and
small mesopore) region. Finally, it is adsorbed at an appropriate site.1
The rate-determining step is the slowest of the steps and will determine
the rate of the whole process. The adsorption or final step is always
fast compared to diffusion (vida supra). For materials with bimodal pore
distributions (like the Rohm and Haas materials) it is convenient to
separate internal pore diffusion into transport pore (macropore and large
mesopore) and micropore (and small mesopore) diffusion. The transport
pores consist of spaces between microspheres (the carbon backbone) and the
micropores are voids in the microspheres (see Figure 1-6).' Since the
pores are so different in size and shape the rates of mass transport
through them are likely to be different and are considered separately
In general, the diffusion rate of adsorbate molecules in solution is
due to the random movement of adsorbate as they are battered around from
collisions with solvent molecules. In a uniform solution, the rate of
random diffusion of adsorbates into any specific region is equal to the
rate in any other region. However, when a concentration gradient exists
then more molecules are available in one region (area of high
concentration) to diffuse towards another (area of low concentration).
This is the situation that exists when a fresh adsorbate granule is placed
in a solution. The solution inside the adsorbent is depleted of adsorbate
due to adsorption onto the surface. The concentration gradient thus
created between the solution outside the adsorbent (high concentration)
and the solution inside the adsorbent (low concentration) causes more
adsorbates to diffuse toward the adsorption sites and fewer to diffuse
away, thus leading to "mass transfer" into the adsorbent.' Diffusion
through stationary liquids can be extremely slow (e.g., a drop of ink in
an undisturbed glass of water). This is overcome either by agitation of
the beads in a batch of solution or by flowing of solution through
No matter how dynamic the mixing in the bulk, a thin film of
stationary solution remains on the outside surface of the adsorbent. The
rate of agitation will determine if the thickness of the film, ranging
from a few molecular dimensions (rapid mixing or flow) to as large as the
container (slow mixing or flow).' When mixing in the bulk is inadequate,
film diffusion will be the rate-limiting step and the adsorption rate will
be dependent on the rate of mixing. As the rate of mixing increases, so
will the rate of adsorption until internal pore diffusion becomes the rate
limiting step. In the case of film diffusion the rate of diffusion is
directly proportional to the surface area of the adsorbent particle and
independent of the nature of the adsorbent.'
The next step in the adsorption pathway in Figure 1-6 is movement
through the transport pores (macro- and large mesopores). Transport
pores, by definition, are large compared to molecular dimensions.11
Therefore, the rate of diffusion through them is often assumed to be the
same as the rate for the bulk, because collision with pore walls will
occur less frequently relative to collisions with solvent molecules.1 Two
other mechanisms have been offered. The first involves the spreading of
the adsorbate along the pore wall (surface diffusion)." The second is
a combination of bulk diffusion and surface diffusion in which the
adsorbate "jumps" through the pore.16 In general the mechanism is not
known and may be different for different adsorbents.1 Whatever the
mechanism, the rate of mass transfer in the transport pores varies
inversely with both the adsorbate concentration and the square of the
particle radius.' Mass transport also decreases with increasing adsorbate
molecular weight19 and increases with increasing macropore void fraction
(volume of macropore/volume of bead).m Part A of Figure 1-7' shows the
Figure 1-7 Adsorbate distribution within a adsorbent for A, transport
pore limited diffusion; B, micropore limited diffusion; C,
the intermediate case of similar transport pore and micropore
relative concentrations of adsorbate in an adsorbent bead when transport
pore diffusion is rate limiting. Initially, adsorbate molecules saturate
the exterior surface of the bead. As the adsorbate front moves through
the bead, a sharp boundary exists between saturated and virgin adsorbent.
Mass transport and adsorption stop abruptly when the boundary reaches the
center of the bead.
Diffusion through the micropores (and small mesopores) is the last
step prior to actual adsorption (vida supra). The mechanisms for
transport pore diffusion are also applicable to micropore diffusion. In
addition, because the pores are on the order of molecular dimensions,11
other processes may slow the rate. Activated diffusion21 can be important
in molecular sieve-sized pores where the adsorption energy is great enough
to trap adsorbate molecules for extended times. Knudsen diffusion21 is
important when collisions with the pore walls occurs at the same rate as
collisions with solvent molecules. In either case the rate of mass
transfer decreases with decreasing pore size. Part B of Figure 1-7
illustrates the case when micropore diffusion is rate limiting. The
concentration of the adsorbate in the transport pores is nearly identical
to the bulk. Macroscopically, the adsorbate is evenly distributed
throughout the bead. Microscopically, a diffusion front progresses within
the microspheres from the surface of each microsphere toward the center.
The amount adsorbed increases as the front progresses as shown in the
figure. When the microspheres become saturated, mass transfer suddenly
In general it is difficult and sometimes impossible to identify the
rate-determining step in the overall adsorption process. Adsorbate
concentration affects the different diffusion rates in a complex manner
such that what may be the rate-determining step at high concentrations may
not be the same for low concentrations. Additionally, the adsorbate's
molecular size and weight influences the diffusion steps differently,
allowing different adsorbents to have different rate-determining steps.
Studies of materials similar to the Rohm and Haas adsorbents in table 1-1
have shown that the distribution of adsorbate in the adsorbent bead is
similar to that depicted in part C of Figure 1-7.1 The rate of diffusion
is more rapid in the transport pores than the diffusion in the micropores.
Initially the rate of adsorption consists of two components, diffusion
through the transport pores and diffusion into those micropores behind the
front. An abrupt change in the adsorption rate occurs when the transport
pore diffusion front reaches the center. The rate of subsequent diffusion
is due to the rate of diffusion in the micropores. Examples due exist
where the rates are identical and no abrupt change in rate is seen.'
Characterization of Carbonaceous Materials
Characterization of these materials is difficult. They are black
insoluble beads or powders that render UV-Vis and IR spectroscopy useless.
Surface area and porosimetry measurements are the most commonly utilized
characterization techniques reported for these materials.11 These
measurements are typically done on samples that are extremely dry and
under vacuum conditions." Though gas uptake experiments are good
indicators for vapor phase use of the adsorbents, the conditions of the
experiment are far removed from the actual conditions where the materials
are used for solution adsorption or catalysis.123'4
The Use of Nuclear Magnetic Resonance (NMR)
In 1962 Gordon reported the observation of two 1H signals for water
on a continuous wave Varian A-60 NMR in the presence of Dowex 50W ion-
exchange resin.' One peak was shown to be exterior water while the other
was interior. In dioxane or acetonitrile no peak attributable to the
adsorption into the interior was seen, though both interior and exterior
peaks were observed when water was added to these solvents.22 Creekmore
and Reilleye found using a Varian HA-100 NMR that the rate of exchange
for water into and out of Dowex 50W (8% DVB cross-linking) was 0.73 s"'.
The Ti for the bulk and interior protons was found to be 2.9 and 0.45 8,
respectively.2* It is important to point out that these resins are of the
gel type and swell in the presence of solvents. They are not the
carbonaceous materials as described above. In particular the Dowex SOW
swell 122, 45 and 56% in water acetonitrile and dioxane, respectively.22
Due to their low cross-linking (2-12%) they would not even be considered
good precursors for pyrolysis. These solids are examples of what are
commonly referred to as "polymer carbons," which are usually useful as
ion-exchange materials but have little use as adsorbents.1
Frankel looked at a series of gels and macroreticular ion-exchange
resins, all of which were products of the Rohm & Haas Co.24 This was the
first study to use materials similar to the precursors of the adsorbent
found in table 1-1. Varian HR-60 and HA-100 'H spectra in water were
obtained to compare the different resins and to determine the cause of the
broadening of the interior and exterior peaks. Five possibilities were
considered to contribute to line broadening of the interior peak: (1)
incomplete averaging (restricted motion) of dipole-dipole interactions
(which is an intrabead effect, and not a homogeneity effect); (2)
heterogeneity effects caused by the distribution of resin bead sizes in
the sample (interbead) effect (i.e., the observed line width is an
envelope of peaks of different chemical shifts for different size beads,
which would be caused if either the degree of sulfonation or water content
was a function of bead size); (3) surface effects due to cracking and
irregularities of the surface; (4) line broadening due to the difference
in diamagnetic susceptibility between the two phases (interior and
exterior); (5) line broadening that reflects a measure of the homogeneity
of the resin interior (intrabead effect), that is, any single resin bead
contains a distribution of pore sizes or configurations and, consequently,
a distribution of chemical environments in its interior structure.2 Based
on his and Gordon's studies,2 which showed the water-line width was
independent of the degree of cross-linking of the gel, effect 1 was ruled
out.24 By separating the beads into different sieve fractions and showing
no difference in either chemical shift or line width, effect 2 was
eliminated.24 Effect 3 was eliminated due to the fact that virgin beads
gave the same spectra as beads that had been fractured via heat stress.24
Temperature dependence studies that indicate increased broadening with
decreasing temperature,224 dependence of the interior line width on
exterior medium (i.e., different solvent),24 and increase of peak
separation with increasing field strength led him to conclude 4 was the
dominant effect. Though he could find no positive test for effect 5, he
ruled out 5 due to the independence on cross-linking and dependence on
outside medium and because materials of grossly different physical
structure (gel vs. macroreticular) gave similar line widths.24 As for the
exterior line broadening, he agreed with Gordon that it was due to
Pearson extended the work of Gordon to look at water in various
hydrated aluminas.25 He found that the NMR line width of the protons was
a function of the surface area as measured by N2 adsorption and that the
fraction of the hydroxyl groups on the surface can be estimated from the
width of the proton NMR line.25 Derouane found using proton NMR that
benzene in pores < 40A underwent freezing point depression in gel
Sternlicht et al. reported one of the first Fourier transform NMR
(FT-NMR) studies, measuring the 13C NMR of amino acids bound to cationic
exchange resins.' Using a home-made 15 MHz NMR they found that the T2
values were shorter than T, and were sensitive to the degree of cross-
linking in the resin while the T, were not. This result is supportive of
effect 5 above.24
The advent of FT-NMR and solid-state NMR has greatly increased the
chemist's ability to analyze both molecules and materials.21 As far as
the study of adsorbents the majority of effort has gone into zeolites and
silicas.29 The last decade has seen the emergence of '129Xe as a viable NMR
probe of solid adsorbents due to its large shift and sensitivity to its
physical environment.2930 Again the majority of the work is with
zeolites2g'030 with only the most recent work on GACs and carbonaceous
adsorbents.3a Smith and coworkers31 have shown that pore size information
of silica gels and porous glasses can be obtained by low-field FT-NMR31A
and magnetic resonance imaging (MRI).31b Ford et al.32 measured the self-
diffusion coefficients of several solvents in polystyrene gels and found
them to be on the order of 104 m2/s with diffusion rates similar to that
reported by Creekmore and Reilly.23 Recent interest in the use of
carbonaceous adsorbents for chemical weapon defense initiated some work by
Beaudry et al.33 studying the 31P NMR of adsorbed nerve agent simulants.
In the work that will be described here, an alternative conventional
high field FT-NMR procedure that yields qualitative measurements of pore
size and distribution at atmospheric conditions of temperature and
pressure has been developed.34 For the first time a high-field FT-NMR
signal of an adsorbate inside a solid adsorbent is resolved into different
contributions from the different pores. Contributions inside the
macropore, mesopore and micropore regions of the solid are observed. Pore
capacity of solids varies with the size of the adsorbate molecule. This
method provides a novel, complimentary technique to N2 porosimetry for
studying solid-adsorbate interactions.
The following chemicals were obtained from either Aldrich, Fischer
or Kodak: CC14, CH2Cl2, CH3CN, C66, (CH30)2(CH3)PO (DMMP), (CH3CH2O)3PO (TEP),
(CH30)3PO (TMP), p-C6H4C12, C,6F6, C6H5CH3, CN. The CC4 was distilled fresh
prior to use. The CH2C12, CH3CN, and CA were distilled and stored over 4
A sieves to prevent water contamination. The remaining probes were
purchased at highest purity and used without further treatment. No NMR or
GC (FID) detectable contaminants were present.
Stock solutions of probe molecules were made. The procedure
involved weighing an amount of probe that would yield the desired molarity
solution into a volumetric and then CC14 was added for the remainder. For
the qualitative runs fresh solutions were made up periodically to ensure
proper concentration progression. For the quantitative runs fresh
solutions were made up prior to each run and concentrations calibrated by
either HPLC (UV) or GC (FID).
Silica Gel (S679-500) was obtained from Fisher. Kieselguhr was
obtained from Matheson Colmen and Bell. Zeolite P powder, lot 9106693K,
and mordenite, lot 9107787K, were obtained from PQ. Zeolite LZ-Y-54
(NaY), lot 13923-55, was obtained from UOP. These materials were used
either "as is" or dried overnight in a vac oven at 100 C.
The following materials were donated by the Rohm and Haas Company:
Ambersorbe 563, lot 2144; Ambersorbe 564, lot 2122; Ambersorbe 572, lot
2125; Ambersorbe 348f, lot 3490; MTV1102, a material prepared by grinding
572 using a ball mill and sieving for material that passed through a 100-
mesh sieve (opening size 150 pm); Kureha carbon bead, obtained from Kureha
Chemical Industry Company, LTD; PLR-0347C, a proprietary support based on
pyrolyzed-polyacrylonitrile. A sample of XE-555 (now called Ambergaurd!
555 by Rohm and Haas Company) was provide by Edgwood Research Development
and Engineering Center (ERDEC), Aberdeen, Md. The activated charcoal
(AC), lot 17104, was obtained from J. T. Baker Chemical Co. An ultra-high
surface area powdered carbon, termed AX21, was obtained from Anderson
Development Company. The materials were used "as is" or treated as
described below. All of these carbon-based materials were rinsed with hot
methanol using a Soxhlet extraction apparatus for 24 hours to remove any
residues associated with processing. The samples were placed in a vac-
oven for 24 hours at 140+ C to remove the methanol and then into capped
vials. Several samples showed a propensity to adsorb water from the
atmosphere. These samples were periodically placed in the vac-oven
overnight to remove any adsorbed water.
Modification of Materials
The HY zeolite was preparedO by ion exchanging NaY with an aqueous
solution of ammonium chloride for 24 hours. This zeolite was then
evacuated at 120 C for 24 hours. It was then soxhiet extracted with
deionized water for 24 hours. This was again evacuated at 120 C for 24
hours to remove water and yield NH4Y. This was followed by drying in a
tube furnace under flowing N2 at 350C for five hours to remove ammonia
yielding HY. Evolution of ammonia was confirmed by a litmus test.
A sample of 564 was placed in flasks with CH3OH. The beads were
intentionally crushed using magnetic stir bars. The process was continued
until all the beads were essentially reduced to powder. The materials
were filtered and dried under vacuum (vida supra).
The chemical modification of the Ambersorbe as well as most any
carbonaceous adsorbent is patented.36 Limited permission was granted by
the Rohm and Haas Company to carry out some chemical modification to the
563, 564 and 572 resins. The procedures described below are similar to
and/or are covered by the patent.m Any reproduction of these procedures
would require similar permission.
Chlorination of the 563, 564, and 572 materials involved the
following procedure. Approximately 5 g of material were placed into 250
ml of 1,2 dichlorethane. This solution was saturated with Cl2 vapors by
bubbling the gas through a gas dispersion frit placed inside the solution.
The solution was refluxed approximately six hours with Cl2 gas added
periodically. After the final addition of Cl2 gas the solution was
refluxed an additional two hours. The solution was then filtered. The
resulting material was washed repeatedly with water and dichloromethane.
When the rinse was no longer acidic the material was dried in the vac oven
(vida supra). Chlorine elemental analysis for the 564 material showed -
4% chlorine and for the 572 material 5% chlorine.
Bromination of the 563 and 572 materials was carried out using the
following procedure. Approximately 5 g of material were placed into 100
ml of chloroform (which also contained a pinch of iron powder) and brought
to reflux temperature. A drop wise addition of 250 ml of a 2.83 M
solution of Br2 in chloroform was made over the course of 12 hours. The
resulting solution was refluxed an additional 6 hours. The solution was
then filtered. The resulting material was washed repeatedly with water
and dichloromethane. When the rinse was no longer acidic the material was
dried in the vac oven (vida supra). Bromine elemental analysis of the 572
material showed 6% bromine.
The 572 material was sulfonated using two different procedures. The
first involved stirring some brominated 572 material in fuming sulfuric
acid for 24 hours. This material was then filtered and rinsed with water.
When the rinse was no longer acidic the material was dried in the vac oven
(vida supra). The second procedure involved stirring some of the 572
material in fuming sulfuric acid for one week. This was then worked up as
The 563 material was fluorinated via an exchange of halides. The
brominated 563 material was stirred in a KF solution in DMSO for 72 hours.
This solution was worked up as above.
The majority of the NMR spectra was obtained on a Varian VXR-300 MHz
machine. Some spectra were also obtained using a Varian XL-200 MHz for
comparison. The TMP solutions were quantified using an SRI 8610-FID gas
chromatograph outfitted with an AT-1000 15 m x 0.54 mm ID capillary column
from Alltech. The acetonitrile solutions were quantified using a HP
5890A-FID gas chromatograph outfitted with an RSL-160 30m x 0.32mm ID
capillary column from Alltech. The benzene solutions were quantified
using a Spectra-Physics HPLC equipped with a UV2000 detector (A = 261),
using a lml/min flow of acetonitrile-water (75:25) mobile phase with a
Sperisorb S5C6 25cm x 4.9 mm ID column from Kontron Analytic.
The general NMR experiment
Approximately 2 cm of solid (for constant volume measurements) or
-0.2 g of solid (for constant mass measurements) were placed inside a
standard NMR tube. Then a solution of probe molecule in carbon
tetrachloride was placed over the solid. The entire sample was then
thoroughly shaken and allowed to settle. The tube was tapped until no
trapped air remained. The tube was then placed in the Varian 300 MHz NMR
such that the solid filled the observed field (see Figure 1-8). The sweep
width was set at 12000 Hz with the placement of 0 ppm set by an external
lock. Either 256 or 512 acquisitions were acquired depending on the
resolution obtained. The temperature of the NMR probe, unless set
otherwise, was approximately 22 C. The 'H, 13C, 31p or '9F NMR of the
solution in and around the solid was observed.
The initial study of any system involved a qualitative concentration
analysis. In general, two or three solutions of < 1 M in probe
concentration were done. In a full qualitative analysis a series of
concentrations were done (0.05, 0.1, 0.2, 0.3, 0.5, 1.0, 2.0, and 3.0 M).
For several systems a quantified study was done. This involved
making fresh solutions for each system. The samples were made up as
described above and the NMR taken. The solutions were then poured off and
the remaining probe concentration determined. The amount of probe
adsorbed was determined by difference.
Figure 1-8 Relative location of sample in the NMR window.
The majority of spectra of the samples were taken within one to two
hours after the solutions were made. These samples are termed "fresh".
In the qualitative analysis, several samples were left in the sealed NMR
tubes and their spectra observed at later times. These are termed "aged"
or "old" solutions. Time periods ranged from 12 hours to one week. There
were several quantitative time studies. For these studies different
samples had to be made to be observed at the separate times. For example,
three different samples of a 0.3M solution would have to be made to
observe the effects of waiting one, two or three days.
Several spectra were obtained at 40 and 60 C. Attempts at higher
temperatures were difficult due to the limitations of the solvent
volatility. Temperature was maintained using the instrument's internal
heater. The temperature was not calibrated by external means.
The T, relaxation times were measured using the standard procedure
as outlined in the VXR-300 user manual. Measurements were made on every
system studied. Measurements were also made on samples of various "ages".
Vapor studies have been carried out on the 563, 564 and 572
materials using CH3CN. The sieves were placed into NMR tubes as in a
normal run. The tubes are then placed inside a larger vial fitted with a
stopcock. This vial is then evacuated for several hours. The stopcock is
then closed and CH3CN is injected into the large vial. The sieves sit in
the presence of the vapor for ca. 24hr. The NMR tubes are then removed,
stoppered and an NMR is taken. A spectrum of adsorbed CH3CN is observed
for all three adsorbents. The 564 sample was allowed to "age" and the
spectrum of the adsorbed CH3CN was observed at later times.
The spectra contain overlapping peaks and it is necessary to resolve
them into their individual components in order to make quantitative
analysis possible. The spectra were digitized using the program OUN-PLOT-
MT. UN-PLOT-IT' is a product of Silk Scientific Inc., P.O. Box 533, Orem,
UT, 84059. The spectra were deconvoluted using Lorentzian curves and a
linear baseline with the program PeakFit~. PeakFit" is a product of Jandel
Scientific Inc., P.O. Box 7005, San Rafael, CA 94912-8919.
Several different types of "blanks" were carried out. One involved
taking the spectra of acetonitrile in the presence of ground up glass from
broken NMR tubes to estimate the broadening caused by the presence of
solid. Several blanks were done of the fresh dried solids in CCl4 without
any probe present. A blank was run were a couple drops of water was added
to a NMR tube containing 572 in CC14 to find were water would appear.
Results and Discussion
A 'H NMR spectrum of 0.3 M CH3CN/CCl4 solution over silica gel
(without any pretreatment) was obtained. A large amount of surface bound
water was observed in the NMR spectrum. A second sample of silica gel was
dried in a vac-oven for 24 hours at 100 C. Figure 1-9 presents the 'H NMR
spectrum of a 0.3 M CH3CN/CCl4 solution over dried silica gel. One peak
is observed that is broadened by the presence of solid in the sample and
by exchange of free and coordinated acetonitrile. A T, experiment on the
same sample results in only one type of environment as indicated by the
inversion recovery. Therefore, it is concluded that this solid does not
function as an adsorbent by micropore filling. The fact that the entire
broad band behaves the same way in the T, experiment suggests that the peak
is broadened by field inhomogeneity and bound molecules are either
I I I
20 10 0 -10 -20
Figure 1-9 'H NMR spectrum of a 0.3 M CH3CN solution in CC14 with dried
exchanging rapidly with CH3CN in solution or are not observed.
A 1H NMR spectrum of a 0.3 M CH3CN/CCl4 solution on kieselguhr (dried
24 hrs, 100 C) is presented in Figure 1-10 and shows what appear to be 2
overlapping peaks. Interestingly, no sharp peak characteristic of "free"
acetonitrile is observed. The down-field peak is slightly larger and
sharper. The 1H NMR spectrum of the sample after it was allowed to
equilibrate 3 days showed only slight changes (not shown). Again there
appears to be two overlapping peaks, however, there is a notable
difference in the distribution of acetonitrile. Curve resolution of the
initial spectra required three curves with the third curve indicating
water was present. Following this result additional spectra were
obtained. The concentrations used were 0.5 M, 1.0 M and 2.0 M (not
shown). As the probe concentration increases there appears to be a single
large broad peak. Again no sharper peak for "free" acetonitrile is
observed even at 2.0 M concentration. T, measurements of the 2.0 M solution
clearly show that all species relax at the same rate of 1.5 sec. The rate
observed is 1.5 s. This rate is about one third that of acetonitrile in
carbon tetrachloride (4.2 s). This data indicates that the "free"
acetonitrile is exchanging with the surface bound acetonitrile. Further,
there are two sites in which the acetonitrile can exchange into. The weak
acidity and low volume of accessible pores indicates that Kieselguhr would
only weakly adsorb material on the surface of the solid.
Constant volume (vida supra) samples of mordenite, P, and NaY were
run with various probes. Figure 1-11 shows an acetonitrile spectrum for
NaY. This spectrum is typical of all three zeolites and is similar to
those found in the literature.2k Curve resolution analysis suggests the
spectrum includes a sharp signal for the exterior acetonitrile at 2.2 ppm
and a more broad peak for the interior signal which is slightly shifted at
t / \
I ~/ \''
j~/ \ 'V
Kh^" / \ *
20 10 0 -10 -20
Figure 1-10 IH MR spectrum of a 0.3 M CH3CN solution in CCl4 with
5 0 -5
Figure 1-11 1H NMR spectra of a 1.0 M CH3CN solution in Cc4 with NaY.
Figure 1-12 1H NMR spectrum of a 0.3 M CH3CN solution in CCl4 with NH4Y.
Figure 1-13 1H spectrum of a 0.3 M CH3CN solution in CCl4 with HY.
Samples of the NH4Y and HY were investigated using a 0.3 M solution
of acetonitrile. In the NH4Y (Figure 1-12) some water was present. The
shoulder on the down-field peak is assigned to water and the down-field
peak (7 ppm) to the NH4. The most intense peak (4.6 ppm) is acetonitrile
and the intensity indicates little adsorption of acetonitrile into the
pores. The spectrum for the HY contains an exceptionally broad peak and
a second peak which is almost broadened into the baseline (Figure 1-13).
The broader peak is probably due to the acetonitrile interacting with the
proton to form a Lewis acid/base adduct which exchanges with excess
acetonitrile in the pores. The narrower peak is attributed to
acetonitrile in solution exchanging with physisorbed acetonitrile on the
exterior of the zeolite. Curve resolution suggests there are four
signals. A broad signal at 6.5 ppm could be either the proton signal or
a H30 species. Peaks at 2 and 1 ppm could be the free and interior
acetonitrile signals, respectively. A second broad signal at -2 ppm may
be the adduct. This peak is exceptionally broad (width at half height =
22.7 ppm) which would be expected if it is mole fraction averaged with the
acetonitrile peak in the pores.
The experiments conducted were designed to elucidate which peaks in
the 'H NMR were caused by probe molecules in the different pore types.
Solutions of various concentrations of benzene, dichloromethane, and
acetonitrile in carbon tetrachloride were used in the analysis. The
solution concentrations used for the qualitative experiments were: 0.05 M,
0.1 M, 0.2 M, 0.3 M, 0.5 M, 1.0 M, 2.0 M and 3.0 M. An interesting
observation was that in general, at low probe molecule concentration, the
564 beads tend to float on the CCl4. As the probe concentration increases,
the beads adsorb the probe and sink to the bottom. The NMR spectra shown
in Figure 1-14 are typical for CH2Cl2. The sharp peak that appears in the
Figure 1-14 IH spectra for 1.0 and 2.0 M CH2C12 in CC14 with 564.
region expected for dichloromethane (4.9 ppm) is assigned as the "free"
probe molecule in the bulk CCl4 solution surrounding the adsorbent. The
next peak up-field (the shoulder at 2.6-2.8 ppm) is assigned to the signal
of the probe in the macropore and large mesopores regions of the
adsorbent. Since these pores are quite large, there would be only a
slight difference between the field felt by molecules in these pores vs.
the bulk solution. Thus, the observed peak is marginally shifted and some
broadening is observed.
The large separate broad up-field peak is in reality two peaks under
a common envelope. Low concentration studies using solutions in the 0.05
M to 0.5 M range show this region of the spectra appearing first. The
spectra also show that the two peaks (the down-field peak, at -4.4 ppm,
first appears as a shoulder on the up-field one, at -8.2 ppm) grow in
intensity until the single broad peak is observed at 0.5 M probe
concentration. T, measurements indicate these up-field peaks relax 2-3
times faster than the others in the spectrum. In the case of
acetonitrile, the T, value for the solvent peak is 2.2 seconds, while the
up-field peaks are 0.7 seconds. These observations lead to the assignment
of these peaks to probe molecules in the micro- and small mesopores of the
More concentrated solutions (1.0 M to 3.0 M) show that the intensity
of the peaks in the up-field region reaches a maximum at about the 1.0 M
concentration while the "free" and macropore resonances increase in
relative intensity with increasing molarity (see Figure 1-14). The
concentration studies show that the probe molecule preferentially fills
the micro- and small mesopores first.
Other observations from the observed spectra can be made.
Porosimetry studies show that there is a bimodal distribution of pore
dimensions (gaps exist in which few or no pores exist with intermediate
sizes). The NMR shows three distinct regions in the observed spectrum,
corresponding to distinct groups of pores with few or none of intermediate
sizes. If there were a continuous distribution in the range of pore
dimensions, a single broad peak would result in the NMR. Further, since
separate peaks are observed, the exchange of probe molecules from one type
of pore to another must be slow on the NMR time scale. Temperature
variation showed little change in the NMR spectrum over the range of 20 to
60 C indicating that exchange makes a minor contribution if any to the
line width. COSY experiments over a seven hour time frame were
inconclusive about exchange between pore signals.
In order to further verify the peak assignments a sample of 564
beads were physically crushed into powder. The macropores which are
expected to be more fragile would be diminished in this processes. The
micro- and small mesopores are more robust and would be affected less.
The resulting NMR spectrum is shown in Figure 1-15. Note that the
shoulder which was assigned to the macropore region is essentially
eliminated. However, the up-field peaks associated with the smaller pores
are still present.
Vapor studies were also used to confirm peak assignments. Any vapor
adsorbed would only be in the smaller pores (vida supra). The spectrum
was taken at three time periods: initial, seven, and nine days. In the
initial spectra there appears to be only one peak. After several days of
sitting in a loosely capped NMR tube the later spectra one can see the
transition of the one peak into two. This is believed to happen as CHCN
evaporates from the sample. T1 measurements on the sample shows times
similar to the up-field peak of the solution samples.
Solution studies as a function of time show some interesting changes
in the spectra. If the samples are allowed to sit for a 2-3 days,
spectral changes occur. An example set of spectra are shown in
Figure 1-16. For samples with low concentrations (5 0.2 M) of probe
molecules the following is observed: the initial spectrum shows nearly all
of the probe molecules are in the up-field region, with some in the "free"
and macropore regions; the "aged" solution shows that only probe molecules
10 0 -10
Figure 1-15 IH NMR spectrum of a 1.0 M CH2Cl2 solution in CC4 on crushed
. I.- ...*o
) 0 -10
Figure 1-16 1H NMR spectra of 2.0 M CH3CN solution in CC14 with 564. The
bottom is a "fresh" sample and the top is a 100 hour old
- "fresh" sample
are in the micro- and mesopore regions with none in the "free" and
macropore regions. For more concentrated solutions, a similar
redistribution with time occurs, but at concentrations greater than 0.3 M,
probe molecule remain in the "free" and macropore region remains. These
changes with time are due to the probe slowly diffusing into the micro-
and mesopores (vida supra). This suggest that the micropore diffusion is
slower than the transport pore diffusion.
A solution containing 0.9 M CH3CN and CH2Cl2 in CC14 was added to 564
adsorbent and the NMR spectrum obtained. Figure 1-17 shows the resultant
spectra along with the spectra of the separate 1.0 M concentration
solutions overlaid on the same scale. The peaks associated with CH2Cl2 and
CH3CN in the 1.0 M solutions are labeled accordingly. The spectrum was
taken within one hour after the sample was prepared. Therefore, the
result reflects a pre-equilibrium distribution (vida supra). The spectrum
suggests that there is no real favoritism for either probe by the
adsorbent at this stage. Curve resolution of the spectrum proved
difficult and the results inconclusive.
The 13C NMR spectra of several of the Ambersorb* 564 samples described
above were obtained. The spectra are less informative but are similar to
those found in the literature.2732 A typical 13C NMR spectrum shows a broad
peak slightly shifted up-field from the solution peak of a probe molecule.
No such shoulder is observed for the adsorbent in neat CC14. The shoulder
is assigned to the probe molecule inside the adsorbent. Attempts at using
"9F NMR using C6F6 as a probe produced a similar broad shoulder.
Solutions of various concentrations of benzene, dichloromethane, and
acetonitrile were used in the analysis. The solution concentrations used
for the qualitative experiments were: 0.05 M, 0.1 M, 0.2 M, 0.3 M, 0.5 M,
1.0 M, 2.0 M and 3.0 M. Spectra for the 0.05 M, 0.2 M and 0.5 M
CH3CN/CCl4 solutions are shown in Figure 1-18. Several observations can
be made from the spectra. The earlier experiments with 564 were designed
,' :: #' ^ \
** * *
10 5 0 -5 -10 -15
Figure 1-17 1H NMR spectra of solutions of 1.0 M CH3CN, 1.0 M CH2Cl2 and 0.9
M in each in CCl4 with 564.
IH NMR spectra for 0.05, 0.2, and 0.5 M CH3CN solutions in CO14
with 572. The bottom spectra is the 0.05 M solution, the top
is the 0.5 M solution.
to assign peaks in the NMR to the different pore types. The middle
spectrum (0.2 M CH3CN/CCl4) in Figure 1-18 should be consulted to follow
the assignments below. The sharp peak (1.6 ppm) that appears in the
region expected for acetonitrile in solution is assigned as the "free"
acetonitrile in the bulk CCl4 solution surrounding the material. In the
figure this is about at 1.6 ppm. The next peak up-field (the shoulder at
1.4 ppm) is assigned to the signal of the probe in the macropore regions
of the sieve. Thus, the observed peak is marginally shifted and
broadening is observed.
The large separate broad up-field peak is in reality three peaks
under a common envelope. Curve resolution analysis of the 0.2 M spectrum
places the peaks at -12.5, -7.5, and -4.5 ppm. The low concentration
studies show this region of the spectra developing first (follow the
development in the figure). They also show that the three peaks grow in
intensity until the single broad peak with a tail is observed at 0.5 M
probe concentration. The spectrum of the 0.5 M CH3CN/CCl4 solution is
shown again in Figure 1-19 to aide in comparisons with other solids, vida
infra. T, measurements indicate these up-field peaks relax 2-3 times
faster (< .4 s) than the others in the spectrum. These observations lead
to the assignment of these peaks to probe molecules in the micro- and
small mesopores. Further, concentration studies using the 1.0 M to 3.0 M
solutions show that the intensity of the peaks in the up-field region
reaches a maximum while the "free" probe and down-field portion increase
with increasing molarity. The concentration studies show that the probe
molecule preferentially fills the micro- and mesopores first.
Vapor studies using 572 yielded results similar to those obtained
using 564. The results of the crushed 572 beads will be discussed below.
The porosimetry data for 572 beads are also similar to the 564 as far as
the bimodality and similar conclusions can be drawn.
The 13C NMR studies were also done. The results are different then
for the 564 adsorbent. In contrast to the 564 adsorbent, 572 always sinks
10 5 0 -5 -10 -15
Figure 1-19 IH MMR spectrum of a 0.5 M CH3CN solution in CCl4 with 572.
in CCl4. Recall that the 572 material is more activated than the 564. The
shoulder expected for the CC14 inside the adsorbent is observed.
For the discussion here, the Ambersorbe 348f was used straight from
the bottle with no pretreatment prior to analysis. Based on the results
obtained for the 564 and 572 materials, concentration studies from 0.05 M
to 1.0 M were conducted for the 348f. The results are similar to those
obtained for 564 and 572, however, some interesting differences are seen.
The 0.5 M CH3CN solution spectrum is shown in Figure 1-20. The up-field
region appears to be only two overlapping peaks. This is similar to what
is observed for the 564 adsorbent. Unlike the other adsorbents, the peak
which eventually dominates is the furthest up-field peak. Curve
resolution suggests that there are actually three peaks (-13.4, -10.3, -
5.9) with the central peak being the dominant one. It is also interesting
to note that the down-field region is more complicated than for the other
materials. There appears to be two regions in the macropore range, until
at 0.5 M, a sharp signal corresponding to the "free" probe molecule first
This material was specially prepared by Rohm and Haas. Ambersorb'
572 material was ground in a ball mill then sieved so that it would be of
the same particle size as that found in the Ambergaurd" 555. For this
study it was also used straight from the bottle. It was expected to have
some adsorbed water. However, the peak is normally removed from those
associated with the probe. Recall, when Ambersorbe 564 was crushed there
was loss of macroporosity when the NMR experiment was carried out, while
the micropore region of the spectra was essentially the same. T,
measurements of the micropore region gave similar results to those of
spectra obtained using the whole bead. The solutions chosen to examine
the MTV1102 were 0.2 M, 0.3 M, and 0.5 M. The 0.5 M acetonitrile solution
spectrum is shown in Figure 1-21. The peak furthest down field is the
5 0 -5 -10 -15
Figure 1-20 'H NMR spectrum of a 0.5 M CH3CN solution in CC14 with 348f.
10 5 0 -5 -10 -15
Figure 1-21 IH NMR Spectrum of a 0.5 M CH3CN solution in CCl4 with MTV.
water peak. The spectrum shows that there is indeed a loss of the
macropore peak. Further the up-field region of the spectra shows a loss
of area. There is no additional growth as was earlier seen for the 572.
Curve resolution analysis suggests that this is due to the loss of much of
the mesoporosity and some loss of micropores.
The 563 sieves have proven to be difficult to work with. They do
not readily adsorb the probe molecules from the solution. However, two
concentration studies using CH2CI2 and CH3CN were done. These two studies
suggest that the 563 more readily adsorbs the CH2Cl2. The studies also
show similar results as those obtained for the 564.
Activated carbon (Baker Food Grade. powdered)
The activated carbon was obtained and dried in a vac-oven at >70 C
for 24 hours for the following study. The solutions chosen for the study
were 0.05 M, 0.1 M, 0.2 M, 0.3 M, 0.5 M, 1.0 M, and 2.0 M. The spectrum
of the 0.5 M acetonitrile solution is shown in Figure 1-22. The spectra
for the 0.05 M and 0.1 M solutions show that essentially all the probe is
adsorbed by the material. From the spectra it is apparent that the
porosity is almost entirely in the micropore and small mesopore region.
Curve resolution analysis suggests peaks at -7.4 and -3.6 ppm for the two
respective regions. Furthermore, no additional probe is picked up by the
solid in solutions with probe concentrations greater than 0.3 M. This is
in contrast to the Ambersorbe materials, which are observed to pick up
probe in solutions t 1.0 M in probe concentration.
Summary of the qcrualitative CH3CN studies
The comparison of the capacity of various carbonaceous adsorbents
for CH3CN is summarized in Figure 1-23. Data is illustrated on the basis
of a constant volume of adsorbent (lighter shading) and on the basis of a
constant mass of adsorbent. Using the latter criterion activated carbon
is seen to have the highest capacity for adsorbing CH3CN from CCl4. On the
Figure 1-22 IH NMR spectrum of a 0.5 M CH3CN solution in CCl4 with AC.
I I 1
572 572(pwd) 348f Act.
Figure 1-23 Acetonitrile adsorption based on the "free" acetonitrile peak
area in the 0.5 M solutions.
* Moles Adsorbed Moles CH3CN/Gram C
basis of the former, 348f is the best. It is interesting to note that the
capacity (for CH3CN) of A572 powder is greater than that of A572 in bead
form. This suggests that small micropores have been made accessible to
acetonitrile by powdering the sample. The amount of acetonitrile adsorbed
was calculated as follows. One ml of a 0.5 M solution of acetonitrile was
placed over 2 cm of the adsorbent (in an NMR tube) and the NMR spectrum
observed within one hour. Curve resolution analysis was performed. The
peak areas associated with the acetonitrile in the bulk ("free") and all
of the pores were normalized to 100%. The amount of acetonitrile
associated with the "free" acetonitrile (ie. the nonadsorbed acetonitrile)
was subtracted from the total amount of acetonitrile (5.0 x 104 moles).
The remainder was considered adsorbed.
Modified Ambersorb* materials
The NMR spectra of all the modified materials were taken using
either CH3CN, CH2Cl2, or C6H6 as a probe molecule. In all cases the spectrum
for the modified material was different from the parent adsorbent. In the
case of the 563-Cl, 563-Br and 563-F materials, the probe molecule was
adsorbed better than the parent 563 adsorbent. The spectra for the 572-
SO3H from direct sulfonation and the 572-SO3H(ex) from exchange appeared
similar. Figure 1-24 shows the comparison for the 564 and 564-Cl
materials. The 564-Cl has a loss of peak area in the small meso-
/micropore region. There is a corresponding gain in peaks associated with
the macropore and "free" benzene.
Quantification of the Method
The prior sections illustrate that the NMR method can be used to
compare different materials. It would be helpful if a more quantitative
approach were available. Curve resolution analysis can provide the
relative amount of probe in the various pore types. A more useful
quantity, however, would be the number of moles of adsorbate in each of
the pore types. This section will describe and give examples of how this
10 0 -10
Figure 1-24 1H NMR spectra for a 1.0 M CH solution in CC14 with 564 and
can be done.
In order to obtain quantified results, fresh solutions of the probe
molecule in CCl4 were made for each series of experiments. A calibration
of the solutions by GC or HPLC was conducted. Solid materials were
carefully weighed out and placed in the NMR tube (ca. 0.2 g). One ml of
a solution of the desired concentration of the probe molecule in CC14 was
placed over the solids. The samples were "aged" for a specified time
(vida infra), and the NMR spectra obtained. As soon as possible after the
NMR spectrum is obtained, the solution above the solid was poured off an
analyzed by GC or HPLC and the concentration of probe molecule in solution
is determined. The amount adsorbed was considered to be the difference
between the amount of adsorbate in the original and final solution. A
best fit curve of the amount of probe adsorbed was calculated based on the
experimental data. The spectra were digitized and deconvoluted. The NMR
peak areas associated with the different pore types were normalized to
100%. The amount of probe in the various pores was determined by
multiplying the normalized peak area by the amount adsorbed based on the
best fit curve.
CH.CN with 572
A set of experiments using CH3CN as a probe with 572 were conducted
as described above. The samples were aged two hours, one day, and three
days. The results are presented graphically in Figure 1-25 and illustrate
the dynamic nature of the adsorption process. In the figure, the moles of
CH3CN adsorbed in the various pores and overall is plotted vs. the
concentration of CH3CN found in the solution above the solid.
Data obtained after a two hour contact time shows that the
adsorption has not reached equilibrium. In the low concentration ranges,
nearly all the adsorption is in the micropores. As the concentration
increases, the amount of CH3CN in the larger pores increases at the
apparent expense of the micropores. The total amount adsorbed appears to
I I i i i I I I
-. .. .._.1 .. -
SI I .
I I I I I "
0..-E-- .- --
I-,- T, , I , I
0.5 1.0 1-5
The amount of CH3CN adsorbed from solutions of various
concentrations into the different pores as determined by NMR
and GC for samples of various ages.
I I I a I
increase with increasing concentration. This type of behavior is expected
if the adsorption rates of the transport pores and micropores are similar
(recall part C of Figure 1-7).
The results of the one day solution suggest that the adsorption
process has not quite reached equilibrium. The micropores are beginning
to level off. There appears to be an increase in the amount of probe in
the macropores as compared to the two hour sample.
At the three day mark the samples have apparently reached
equilibrium. Total adsorption increases and then levels off. For the
lowest concentration solution, the majority of the adsorbate is in the
micropores. For the remaining solutions the micro- and mesopores appear
to have a constant amount of adsorbate. The increase appears to be in the
macropore region which now has an appreciable concentration of probe at
the lowest concentration solutions.
C.9 with 572
A series of solutions was also studied using benzene as a probe.
The samples were aged one day, the NMR spectra obtained, and the solutions
analyzed by HPLC. The results are presented in Figure 1-26. The results
suggest that the adsorbent has not reached its capacity for benzene at the
highest concentration studied. The amount of benzene in the micropores is
constant throughout the range studied. The amount in the meso- and
macropores increases with increasing solution concentration.
Some interesting comparisons can be made between the one day
solutions of CH3CN and C6H6. Recall from Figure 1-5 that the forces
affecting adsorption are attraction of the solvent for the adsorbent, the
attraction of the solute for the adsorbent, and the solubility of the
solute in the solvent. Recall further that the AH, can be used as a
measure for the attraction of solute with the adsorbent surface. Benzene
and acetonitrile have the same AH (7.3).37 The solvent is CC14 and has
been held constant. Therefore the only difference is the interaction of
acetonitrile and benzene with CCd4. A parameter which is often used to
Figure 1-26 The amount of C^,H adsorbed in the various pores as determined
by NMR and HPLC for a one day sample of various solutions.
I I I I I I I I I I I I I I I I I I I I I I I I
*11 I I li I 'i'' I I I lii
: .... o"-*^
0-."" E 9--- l E
0t I I I
qualitatively understand the solubility of mixed solutions is the
Hildebrand number, 6.3 Loosely speaking, the closer the 6 values of the
two liquids the greater the solubility of the one is in the other. The 6
values for carbon tetrachloride, benzene, and acetonitrile are 8.6, 9.2
and 11.2, respectively. These values suggest that the reason more
acetonitrile is adsorbed vs. benzene is because benzene is more soluble in
carbon tetrachloride, which is consistent with other adsorption studies. 3"
For both systems the probe has a greater affinity for smaller pores
over the transport pores. This is evident from the low concentration
solutions. The affinity for the small pores is consistent with the
physical effects of adsorption discussed in the introduction. The NMR
results are consistent with gas uptake measurements.39
Chemical Warfare Agent Simulants
Carbonaceous materials and GACs have been used for chemical weapons
defense for several years.40 The current solid sorbent system used as the
primary skin decontaminant is the M291 kit. The kit is composed of
nonwoven fiber pads that are filled with a resin mixture developed by Rohm
& Haas (trade name: Ambergaurdo 555). Ambergaurd! 555 is composed of a
three part mixture of an adsorbent (crushed Ambersorb 348f), a strong base
resin (crushed Amberlite. IRA-900), and a strong acid resin (crushed
Amberlyst* XN-1010). An NMR investigation of the Ambergaurd! 555 resin
showed the first direct evidence for agent-resin interactions." It showed
that GD (3,3-Dimethyl-2-butylmethylphosphonoflouridate) hydrolyze with a
half life of 10 days on the basic component of the resin. However,
neither VX (0-Ethyl S-2-(diisopropylamino)ethyl methylphosphonothiolate)
nor a HD simulant (13CH3SCH2CH2Cl) hydrolyzed during the first 10 days. It
was determined that the adsorbent component provided the majority of the
adsorption. Of the two exchange resins, which were supposed to provide
catalytic decomposition of the agent, only the strong base resin
decomposed the simulant.3 Dimethyl methylphosphonate (DMMP) was used as
a simulant for the "G" type nerve agents for much of the NMR work.
Since the purpose of the decon kit is to either remove or destroy
the agent in a rapid manner, the study by Beaudry et al.,33 illustrated
that the solid adsorbent is the primary material in accomplishing this
objective. In the sections below the solution high-field FT-NMR method
will be used to investigate several solid adsorbents using "G" agent
The following solids were chosen for study by NMR with DMMP as the
probe: Ambersorb! 572, Ambersorb* 348f, MTV 1102, and activated carbon.
In order to test if the method would work with 31P NMR, a 1.0 M solution
of triethyl phosphate (TEP) in CCl4 was used with 572. The 31P NMR spectrum
of this molecule in solution without any solid adsorbent present is a
singlet. The spectrum of the mixture describe above (Figure 1-27) was a
slightly broadened "free" probe peak with two additional shoulders of
different intensities. The 1H NMR of the same sample (not shown) confirmed
that some probe had been adsorbed by the solid.
The concentration of the DMMP for the next studies was 0.3 M. The
first material attempted was Ambersorbe 572. The 31P NMR spectrum obtained
(Figure 1-28) was not as informative as the test spectrum. The peak
observed is broad. There is evidence of a shoulder slightly up-field.
However, there is little hope for obtaining meaningful information from
As noted earlier, the 1H NMR of TEP was used to confirm that the
adsorbent had adsorbed the probe. A proton spectrum was run on the above
sample (not shown but a similar spectra is shown vida infra) and confirmed
that the DMMP had been picked up. Therefore, it was decided to
investigate the DMMP adsorption using 1H NMR. It should be noted here that
DMMP contains two magnetically inequivalent proton sites. Therefore, the
proton spectrum is not as easy to interpret as for acetonitrile.
Figure 1-27 The 31P NMR spectrum of a 1.0 M TEP solution in CCl4 with 572.
40 20 0
Figure 1-28 The 31p NMR spectrum of a 0.3 M DMMP solution in CC4 with 572.
^/ ^ ^^^^ 572
5 0 -5 -10 -15
Figure 1-29 The 1H NMR spectra of 0.3 M DMMP solution in CC14. with (from
bottom to top) 572, 348f, AC, and MTV 1102.
Fresh solutions of the materials mentioned above with the 0.3 M DMMP
solution were made. The proton spectra were obtained and the results can
be seen in Figure 1-29. Interestingly, the Ambersorbe materials appear to
have a higher adsorption capacity than the activated carbon (compare the
bottom two spectra with the second). Furthermore, the best material
appears to be 572. Another important result can be seen by comparing 572
(bottom of figure) with 1102 (top spectrum). As stated earlier, 1102 is
simply crushed up 572. The adsorptive capacity of the 1102 material is
greatly reduced compared to its parent material under these conditions.
This suggests that microporosity may not be the only important feature for
adsorption. If this is true, the adsorption ability of the crushed up
348f found in Ambergaurd* 555 is reduced over what could be obtained with
larger beads. Spectra taken a couple days later (not shown) show some
additional pick up and redistribution in the 348f and 572 materials.
Due to the multiple proton signal of both TEP and DMMP, and the
desire to quantify results, trimethyl phosphate (TMP) was used as a probe.
TMP has a singlet in the 1H NMR in the absence of a solid adsorbent. A
qualitative run of solution concentrations was done to find the best range
for the quantitative experiment. The spectra for the 0.2, 0.4, 0.5, 0.8,
and 1.0 M TMP/CCl4 solutions are shown in Figure 1-30. These spectra were
obtained using an NMR window of 14,000 Hz. Curve resolution suggests the
following peak assignment. The free TMP is at 1.9 ppm. The macropores
are assigned to a peak at 3.2 ppm. The down-field region is composed of
two peaks at -5.9 and -16.3 ppm.
For the quantitative studies the concentrations used were 0.1, 0.2,
0.3, 0.4, 0.6, 0.8, and 1.2 M TMP in CC14. A sample of both 572 and 564
with each of these solutions was made and allowed to age one day. The NMR
spectra were obtained and the solutions analyzed by GC. The results are
presented in Figure 1-31.
_______0 0.4 M^
10 5 0 -5 -10 -15
Figure 1-30 The IH NMR spectra of several solutions of TMP in CCl4 with
0.2 0.4 0.6 0.8 1.0 1.2
[TMP]soil, 1 day
0.1 0.2 0.3 0.4 0.5 0.6
TMP]sol, 1 day
Figure 1-31 Results of quantitative NMR studies of TMP solutions with 572
(top) and 564 (bottom).
The 572 adsorbent, with its larger pore volume adsorbs more TMP.
Both adsorbents reach a maximum amount adsorbed and level off. It can
also be concluded that the system has not reached equilibrium. For the
572 system the amount of TMP in the micropores is increasing as the
overall concentration increases. The effect is similar but less dramatic
for the other two pores.
The 564 system proved interesting. As noted earlier in the
quantitative studies of other probes, 564 will float on CC4 until enough
probe molecule is adsorbed to cause the beads to sink. When the seven
solutions were lined up side by side, the amount of beads on the bottom of
the NMR tube increased with the increasing probe concentration, giving a
qualitative assessment of the adsorption. Once again there are two peaks
in the down-field region at -6.7 and -17.1 ppm. The free TMP is at 1.8
ppm and TMP in the macropores is at 3.5 ppm.
The major difference between 572 and 564 is the amount of
microporosity. This appears to account for the difference in the amount
of TMP adsorbed. Where the concentration of TMP in the micropores in 564
is the same throughout the range, the amount of TMP in the micropores of
572 increases with increasing concentration. The amount of TMP in the
mesopores increases with increasing concentration for both solids.
According to the values in Table 1-1 the amount of macropores is similar
for both adsorbents. However, the NMR results suggest that the amount of
TMP in the macropores of the 572 increases with increasing concentration.
The amount of TMP in the macropores of 564 remains the same.
The above results again suggest that the diffusion in the transport
pores is faster than in the micropore. Further, as with the DMMP, the
ability of the transport pore to act as a reservoir to "store" the
adsorbate until such time as the micropores are capable of diffusing the
adsorbate to an adsorption site is an important part of the overall
process of adsorption of the TMP. Therefore, when it comes to larger
adsorbate molecules, the availability of macropores increases the ability
of the solid to remove the adsorbate from solution.
EXTENSION OF THE UNIFIED SOLVATION MODEL
TO ACCEPTOR SOLVENTS AND THE INTERPRETATION
OF SOLVENT CONTROLLED REACTIONS
The pronounced influence that solvents have in many areas of
chemistry, e.g. reaction rates and physiochemical properties, have
prompted extensive studies aimed at producing a scale of solvent polarity.
Solvation has been subdivided into specific and non-specific effects by
some workers41. Specific effects include localized donor-acceptor
interactions involving specific orbitals. These interactions have been
successfully correlated, predicted and understood with the E and C
equation.42 Non-specific effects involve the interactions modelled by the
reaction field or Kirkwood approaches.43 Solvent reorganization occurs
to form a cavity that accommodates the solute with stabilization resulting
from the interaction of the solute dipole (and induced dipole) with the
internal dielectric constant of the cavity. Solvent rearrangement and
induced dipole moments tend to create an internal dielectric different
from the bulk dielectric constant. In view of the difficulty in
determining the radius of the cavity formed and the internal dielectric of
the organized solvent region, the quantitative application of these models
to the interpretation of solvent effects is not possible without making
assumptions. Accordingly, empirical approaches have been employed to
arrive at solvent polarity parameters to describe solvent
A large number of scales of solvent polarity have been offered in
the literature.'53 Most differ in significant ways so an investigator
correlating complex chemical phenomena can usually find one that works.
Unfortunately, this approach provides little insight about the system,
e.g. why don't the other scales work. One reason for the diversity of
scales arises from the fact that all reported studies include both
specific and non-specific solvation effects. If one parameter can
incorporate both effects, there would be no need to have two effects.
In one of the more recent approaches three solvent parameters, which
involve different physiochemical properties, are offered to treat
The complication of one parameter incorporating both specific and
non-specific interactions can be described by an example using Reichardt's
ET(30) scale of solvent polarity.45 This exhaustively studied and most
widely used scale is based on the negative solvatochromism of pyridinium
N-phenoxide betaine dye, betaine (Figure 2-1), and it has been used as a
measure of the solvent polarity by Reichardt45'4 and others.- Reichardt
and coworkers,56 have even considered the sensitivity of betaine to
acidity in the case of acidic alcohols like CF3CHPOH and to steric effects
in the case of bulky alcohols like (CH3)3COH. Nevertheless, it is assumed
that ET(30) measures essentially the solvent polarity. Work by Catal&n and
coworkers57 has shown that the extremely strong negative solvatochromic
effects on the stilbazolium betaine, SB (Figure 2-2), are not do to the
polarity-polarizabilty effect of the alcoholic solvent but to its
hydrogen-bond acidity. This was proved by studying the effects of the
same solvents on a betaine where the basic center, the oxygen atom, was
protected against the approach of the solvent by one or two bulky t-butyl
groups in the ortho position, TBSB and DTBSB respectively in Figure 2-2.
When this specific solvent effect is prevented, stilbazolium betaines show
a very weak solvatochromism but present a coarse structure in their
electronic spectra due to hydrogen bonding; vice versa, the appearance of
Figure 2-1 Pyridinium N-phenoxide Betaine Dye
Figure 2-2 Stilbazolium Betaine and its Derivatives
a resolved maximum indicates steric hinderance for the approach of an
acidic solvent." The authors felt that the stilbazolium betaines would
better illustrate the effect discussed above because betaine is not so
well protected against solvation effects by the phenol rings since they
can rotate, i.e., the steric effect of a phenyl ring is conformationally
dependent.57 Thus, some solvents that would be hindered by DBTSB would not
be hindered by betaine.
In work described previously, a common single parameter scale of
solvent polarities was found5' which incorporates data from most of the
literature systems by excluding specific interactions. The resulting
parameters5' allow one to predict non-specific solvation interactions for
a wide variety of solutes and solvents.
The selection of systems that do not involve specific donor-acceptor
interactions in the measurement is difficult because of our incomplete
understanding of weak interactions. One approach that has been used to
detect subtle, specific interactions in solvents is referred to as ESP
(Elimination of Solvation Procedure)." A series of reactions:
B + A-S -> AB + S (2-1)
Bi + A*S > AB, + S (2-2)
is studied in poorly solvating media and then in basic, slightly polar
solvents, S. The symbol A*S, indicates that the acid is completed by the
solvent. Subtracting Equation (2-2) from (2-1) leads to:
B + AB. > AB + Bi (2-3)
The specific interaction of A with the solvent has been subtracted out of
Equation (2-3). Providing that B and Bi do not undergo specific donor-
acceptor interactions with the solvent, only non-specific solvation
remains. The experimental enthalpy of reaction for Equation (2-3), for a
given base pair, is a constant" in a wide variety of solvents in which
only non-specific interactions exist. Since dispersion interactions are
a function of the molecular weight of the solute, non-specific solvation
of the products equals that of the reactants and cancels.
An interesting result is obtained when the solvent is varied and the
system does not produce the constant enthalpy expected for Equation (2-3).
This finding indicates that either specific interactions between Bi and the
donor solvent exist or non-specific solvation enthalpies of the product
and reactant fail to cancel. In this manner, specific interactions and
unusual non-specific solvation are detected. Carbon tetrachloride forms
weak adducts (-lkcal mole"') with donors that have large CB numbers, e.g.
nitrogen or sulfur donors. Charge transfer complexes involving pi-donor
and pi-acceptor interactions are observed41c between solvents with pi-
systems e.g. CH6, or o-C12CH4 and solutes with ir-systems. There is
evidence to suggest that these donor-acceptor interactions involve
pyridine with benzene and o-dichlorobenzene and even occur between
pyridine molecules in liquid pyridine.60 These studies-9' also show that
non-specific solvation of the products and reactants do not cancel when
1,2-dichloroethane is used as a solvent. Systems that are well behaved in
o-dichlorobenzene are not when 1,2-dichloroethane is used as a solvent.
The non-specific solvating properties of 1,2-dichloroethane are
complicated by shifts in the equilibrium that exists between staggered and
eclipsed forms of this solvent molecule when it solvates. This solvent is
also capable of forming hydrogen bonds to donor solutes. It is best to
avoid 1,2-dichloroethane for the quantitative determination of solvent
effects. Keeping the above points in mind, systems are selected from
extensive literature data to develop a scale of solvent polarities.
The Universal Solvation Model
In an earlier articles, a scale of solvent polarity was presented
which enables one to estimate the influence of non-specific solvation on
a wide variety of physicochemical properties for solutes of widely varying
shapes and polarity. The equation used to treat non-specific solvation is
X = S'P + W (2-4)
where X is the value of the physicochemical property measured in the
specified solvent; S' is a measure of the solvent's polarity; P is a
measure of the susceptibility of the solute probe to solvation; and W is
the value of X at S' equals zero. The S' values provide a scale of non-
specific solvating ability. Substitution of a solvent's and a probe's
parameters into Equation (2-4) produces the value of the probe property
observed in that solvent. Care was taken to exclude from the data set any
systems in which there were contributions from specific donor-acceptor
interactions. Donor probes are only measured in donor solvents, and data
for acceptor or donor pi-solutes measured in pi-solvents are excluded. By
eliminating specific interactions, all of the experimental data which, in
the past, were used as the basis for several different scales of solvent
polarity were found to be consistent with the one, new unified scale. The
few exceptions involved measurements of polar probes in non-polar solvents
and non-polar probes in polar solvents. These combinations lead to
aggregation of the probe resulting in a molecular environment for the
probe that is not entirely solvent. Exceptions to the model are also
anticipated when a probe is studied whose dimensions are smaller than the
dimensions of the cavities that can be created in the pure solvent.
Ineffective solvation of the solute occurs in these circumstances.
With Equation (2-4), we are in a position to predict non-specific
solvation influences. Equation (2-5)
-AX = E^EB + C^CB + W (2-5)
has been used42 to correlate the donor-acceptor contribution of a variety
of physicochemical properties in poorly solvating solvents where non-
specific salvation contributions are minimal. Recently, the solvation
model was extended61 to systems in which the property AX involves probes
that are acceptors involved in specific donor-acceptor, hydrogen bonding
interactions in polar solvents. For example, changes in the electronic
transition of the acceptor probe 4-nitroaniline were studied" in neat,
polar donor solvents. The NH2 group of aniline hydrogen bonds to the basic
solvent," and the adduct formed is non-specifically solvated by the polar
solvent. For this situation, Equation (2-4) and Equation (2-5) are
combined to accommodate both non-specific and specific interactions.
Equation (2-6) results.
AX = EAEB + CA*CB + S'P + W (2-6)
The EB and CB parameters for the solvent are those reported42 for these
donors reacting with a wide range of acceptors in poorly solvating
solvents. The successful fit of physicochemical data for acidic probes in
neat, polar donor solvents with Equation (2-6) is reported1 The use of
Equation (2-6) and the reported solvent parameters for the analysis of
physicochemical measurements on new acceptor solutes in polar donor
solvents which both coordinate and non-specifically solvate the acceptor
solutes is described.61 Analyses of the data sets used" to establish the
,-wr parameters using Equations (2-4) and (2-6) indicate61 that these
systems have complications from both r-I charge transfer interactions and
incomplete complexation of the solute. These effects are averaged into
the derived P and x* parameters and limit their applicability.
In an extension of the Unified Solvation Model (USM) the very
important class of polar hydrogen bonding solvents is added. Since these
solvents are capable of undergoing both non-specific and specific donor-
acceptor interactions with donor solute probes, the relevant equation is
AX = EA'EB* + CA'CB* + S'P + W (2-7)
The prime values denote parameters that are consistent with the enthalpy
based parameters of the ECW model42 but are determined in the neat acceptor
as the solvent. The specific interaction parameters of the neat solvent
may differ slightly in some instances from parameters for the acceptor in
the gas phase or in a dilute poorly coordinating solvent. For example,
self-association of the acceptor in the pure acceptor solyent could lead
to different parameters for the large aggregate than for the monomer or
smaller aggregate in the gas phase or poorly solvating solvent. When
uncertainty exists about the transferability of parameters measured in
pure solvents to studies in the gas phase or in poorly solvating solvents,
the prime symbol will be employed. It is to be emphasized that the eA' and
CA' parameters are consistent with the enthalpy based parameters we have
reported,42 have units of (kcal mole'1)12, but are to be used to treat
specific interactions in the pure acceptor as the solvent.
This work provides the basis for extending the unified scale of
solvent polarity to hydrogen bonding, polar acceptor solvents.
Experiments are reported, and a set of E^', CA' and S' values are given,
which, for the first time, enables one to determine non-specific and
specific solvation components of the solvation of solutes in hydrogen
Results and Discussion
Systems Involving Non-Specific Interactions
Equation (2-7), for acceptor solvents, contains a large number of
unknown quantities. In order to facilitate finding the minimum for the
best set of parameters to fit the solvent shift data, probes are employed
for which P and W can be determined independently in non-specific
solvating solvents. Donor probes are studied in donor solvents,5 and
these data are treated separately with Equation (2-4) to determine P and
W. Analysis of data for a probe whose P and W values are known in
acceptor solvents leaves (in Equation (2-7)) EB* and C.* to be determined
for the probe and EA', CA' and S' to be determined for the solvents. The
addition of new probes led us to refit the data set previously reported-
for non-specific solvation of donor probes in donor solvents. Data for
366 spectral shifts lead to 366 simultaneous equations that are solved for
34 S' values and 82 probe parameters. The refined S' and P values are
given in Tables 2-1 and 2-2, respectively. The agreement between the
experimental shifts and the shifts calculated by substituting the
parameters from Tables 2-1 and 2-2 into Equation (2-4) is comparable to
that reported earlier." The complete fit is available in the appendix.
Table 2-1 lists the solvent polarity parameters for donor solvents
under conditions where specific interactions with the probe are not
involved. In some solvents, only a limited number of well-established
probes have been studied, leading to tentative S' values. These solvents
are also listed in Table 2-1, and the limited probes used in their
determination are given in the footnote.
Table 2-2 lists the probe intercept (W) and susceptibility (P)
values for 41 different probes. Abbreviations that are used for these
probes in the discussion and computer fits are indicated in parentheses.
Combining the probe parameters with S' in Equation (2-4) enables one to
calculate the spectral shift of the probe from non-specific solvation.
The average absolute deviations, x, of the various probes in the data fit
are given in the footnotes to Table 2-2. The % fit gives the average
deviation as a percentage of the range of shifts observed for the probe.4'
The S' values of new solvents, which only non-specifically solvate
the probes, can be determined and added to the correlation by measuring
the shifts of several probes in the solvent. A series of equations of the
form of Equation (2-4) is written for each probe and solved for S'.
Alternatively, AX-W can be plotted vs P, and the slope of the least
squares line will give S'.
New probes can be added or physicochemical data can be analyzed for
non-specific solvation contributions by measuring AX in a series of non-
coordinating solvents. The series of equations of the form of Equation
(2-4) is solved for P and W. A good fit of the data indicates that the
measured changes of these probes with solvent variation are caused by non-
Table 2-1 S' Parameters for Solvents
No. Solvent S' No. Solvent S'
1 C6Hi2 1.11 24 CHC (0) OCHI' (2.35)"
2 (CH5) 3N 1.43 25 CH3C(O)C;2H5 2.51
3 CC14 b 1.49 26 CjNH 2.44
4 CS2 1.51 27 (C2Hp)3PO 2.55
5 (n-C4H.)20 1.58 28 CH5C (0)CH 3b 2.52
6 C6H5CH3' 1.66 29 C6HSCN" 2.63
7 C6Ha 1.73 30 CJHN025 2.61
8 (CH) O20 1.73 31 (CH3) 2CO 2.58
9 (CH)4S (1.83)d 32 [ (CH3)2N]3PO 2.52
10 C12C=CHClbc (1.90)f 33 CH3CON(CH3)2 2.70
11 C13C-CH3 ',c (1.93)1 34 CH2CH2CH2CONCH3 2.62
12 C6HSN (CH3) (1.96)d 35 [ (CH3)2N]2COC (2.48)d
13 (CH2)4SC (1.99)d 36 C2H5NOc (2.78)d
14 0(CH2CH2) 20 1.93 37 C2H5CN (2.80)'1
15 CHOCH 2.04 38 (CH3) 2NCNC (2.81)4
16 CIsH5Cl 2.07 39 (CH30) PO) (2.79)*
17 (CH2) 5 1.98 40 HCON(CH3)2 2.80
18 (CH2) 40 2.08 41 4-Butyrolactone 2.86
19 l, 2-C12C6H4b'c (2.13)4 42 (CH2)4S012 (2.88)-
20 CH3C (O)0 C2H5 2.15 43 (CH3) 2SO 3.00
21 Quinoline'A (2.30)d 44 CH3CN 3.00
22 (n-C4HO0) 3POc (2.30)a 45 CH3NO2 3.07
23 (CH )jCO 2.35 46 (CH2)3(0-)2CO 3.10
(a) w-acceptor solutes must be avoided.
(b) Strong donor nitrogen, sulfur and phosphorus solutes must be
(c) Limited data is available on these solvents so an n-value of 1 is
used in data fits compared to 0.2 for established solvents.
(d) Not included in fit and calculated from (ET(30) 19.63)/8.61.
(e) Not included in fit and average value calculate from ET(30) and (631P
(f) Not included in fit and average ET(30) and (Michler's ketone -
P and W Parameters for Probes
(Tentative values qiven in parenthesis)
Probe (Symbol)' P W
v; -N,N-diethyl-4-nitroaniline (NNE4NO2AN) -1.69 29.31 0.22
v; -N,N-dimethyl-2-nitroanilineb (NNM2NO2AN) -0.99 26.19 0.17
v-; -NN-diethyl-3-methyl-4-nitroaniline -1.55 29.16 0.19
v;-N,N-dimethyl-2-nitrotolueneb (NNM2NO2TOL) -0.95 25.60 0.13
v ;-4-nitroanisole (4NO2ANISOL)' -1.29 35.51 0.18
av;-4-(2,4,6-triphenyl-l-pyridinio)-2,6- 8.61 19.63 0.20
diphenylphenoxide4 (BETAINE) _____ _____ ___
P;Bis 2-(2-pyridylbenzylidine-3,4- 1.66 11.69 0.11
dimethylaniline,biscyano iron(II))* (Burgess) }
619F;1,4-difluorobenzenef (F2C6H4) -0.36 7.26 0.11
619F; 1-fluoro-4-trifluoromethylbenzenef (CF3C6H4F) 0.39 4.72 0.13
619F; 1-fluoro-4-nitrobenzenef (NO2C6H4F) 0.59 8.61 0.12
619SF; 1-cyano-4-fluorobenzener (CNC6H4F) 0.49 8.42 0.08
615N; 1-methylsilatrane-N(CH2CH20)3SiCH3 (N15)' 4.39 -4.88 0.22
v; 1-ethyl-4-methoxycarbonylpyridinium iodide(Z- 13.23 31.38 0.20
AN; di-t-butyl nitroxide' (ANTBUNO) 0.240 13.967 0.09
v; N,N-(dimethyl)thiobenzamide-S-oxide 1.27 78.28 0.25
AN; 4-amino-2,2,6,6-tetramethyl piperidine-1-oxyl 0.229 14.072 0.09
v; a-[4-(N,N-dimethylamino)phenyl]imino- -2.41 73.91 0.26
acetoacetanilide) (Me2NC6H4NCR2) ______ ______
v; Pyridine-N-oxide (NUPYNO)' 0.36 35.00 0.15
ivE; 1-methyl-4-cyanoformylpyridinium oximate 3.90 39.46 0.34
(OXIMATO-B)" _____ ______
Brookers IVO 8.42 28.24 0.24
613C; N,N diethylbenzamide 0.92 29.79 0.12
613C; C5H5N 0.89 ______ 0.16
613C; CH5NOq 1.92 -19.80 0.15
Tmoauinolinium Ylide TSQOOUTN-YLTD1 13.06 1.92 1.03
Table 2-2 -continued
IProbe (Symbol) P W n
61N; Pyridine-N-oxide (N(14)PYNO)' 5.29 69.53 [0.46
619F; l-fluoro-4-fluosulfurylbenzene 0.79 11.20 0.19
(FC6H4SO2F)t I ____
619F; 1-fluoro-4-pentafluosulfurbenzene 0.51 4.90 0.16
AE(s->t); Nickel-N,N1-di-(p-tolyl) -0.60 4.10 0.13
aminotroponeimineate (NiAmtrop)t I I I______
v; Ni(II) bistrifluomethyldithiolene-1,10- 4.27 45.55 0.22
phenthroline (Ni(tfd)Phen)" ______ ____
631P; triethylphosphine oxide (31P(C2H5)3PO)v 5.09 -8.91 0.29
vE2; 1-methyl-4-cyanoformylpyridinium Oxidmate 3.10 65.77 0.44
(Z')h 14.65 23.96 0.46
v(fl); 7-amino-4-methylcoumarin (COUM)w -1.45 27.07 0.26
v(fl); 7-N,N-dimethylamino-4-methylcoumarinw -1.43 27.91 0.36
Brownstein's S' Parameter' S'(bat) 0.090 -0.392 0.08
1-C2Hs5-4-NO2C2C4 -0.99 38.57 0.20
vi-4,4'-bis(dimethylamino)benzophenone -1.18 31.38 0.20
(Michlers Ketone)y' ______
v-cis-dicyanobis(1-10)phenanthroline Iron 1.38 12.49 0.14
(II), Burgess (Sp)z
(a) The electronic transition energy in kK (lkK = 1000 cm'1). Data from
reference 44b. x is 0.16 and % fit = 5.8 for NNE4NO2AN and 0.11 and
4.5% for 4NO2ANISOL. Estimated experimental error is 0.1 kK.
(b) The transition energy in kK. Data from reference 44b. x = 0.10 and
the % fit is 5.0 for NNM2NO2AN, 0.13 and 6.6% for NNM2NO2TOL and 0.13
and 6.7 for 1-ethyl-4-nitrobenzene. Estimated experimental error is
(c) Transition energy in kK. Data from reference 44. x omitting
acetone = 0.12 and the % fit = 3.6 and the experimental error is 0.1
(d) Transition energy v kcal mole1. Data from reference 45. The v
value in hexane is 30.9, the x = 0.14 and the % fit = 1.
Table 2-2 -continued
(e) Parameters to calculate v kK. Data from reference 46 and 62. The
0.04, the % fit = 1.9% and the experimental error is 0.1 kK.
(f) The 19F chemical shift in ppm relative to fluorobenzene as an
internal standard, x = 0.04 and % fit 6.2%. x 0.06 and % fit
= 7.5; x = 0.05 and % fit 3.8; x = 0.02 and % fit 2.4; x = .12
and % fit = 8.7; x 0.9 and % fit = 10.5 for the F, CF3, NO2, CN,
SO2F and SF5 derivatives, respectively. The experimental error is
0.08 ppm. Data from reference 48.
(g) The 31 chemical shift in ppm for 1-methylsilatrane relative to
cyclohexane. Data from reference 63, x = 0.16 and % fit = 2.0.
(h) Transition energy in kcal mole-. Data from reference 64. In most
instances, the transition is concentration dependent and has been
extrapolated to zero solute concentration. The x = 0.14 and the %
fit = 1.1; x = .17 and % fit = 2.8.
(i) The nitrogen hyperfine coupling constant in cm"1 x 104. Data from
reference 65 where AN is reported as the line separation in gauss
which is actually AN/gP. Since g is not given, it is assumed to be
2.0047 and P = 4.6686 x 105 cnm'/G. Multiplying the line separation
by 9.3591 x 105 gives AN in units of cm-1 x 104. The fit is run by
multiplying the numbers by 104. The x = 0.03 and the % fit = 6.7 and
0.03 and 7.1. Estimated experimental error is 0.01 x 104 cm"1.
(j) Transition energy in kcal mole-1. Data from reference 66. The x
0.21 and the % fit = 8.7.
(k) Transition energy in kcal mole-1. Data from reference 67. The x
= 0.23 and the % fit = 5.5.
(1) Transition energy in kK. Data from reference 68. The x 0.08
and the % fit = 8.9. Transition energy is 0.1 kK.
(m) Transition energy in kcal mole'. Data from Reference 69. x = 0.1
and % fit = 5.7; x = 0.17 and % fit = 2.8.
(n) Transition energy in kcal mole"1. Data from Reference 9. x = 0.19
and % fit = 3.8.
(o0) Chemical shift in units of ppm. Data from Reference 70. x = 0.05
and % fit = 5.7.
(p) Chemical shift in units of ppm. Data from Reference 71. x = 0.02
and % fit = 1.6.
(q) Chemical shift in units of ppm. Data from References 71 and 72.
x = 0.08 and % fit = 6.3.
(r) Transition energy in kcal mole"1. Data from Reference 73. x =
0.003 and % fit = 0.1.
Table 2-2 -continued
(s) Chemical shift in units of ppm. Data from Reference 74. x = 0.17
and % fit = 2.5.
(t) Chemical shift in units of ppm. Data from Reference 75. x 0.06
and % fit = 6.7.
(u) Transition energy in kcal mole1. Data from Reference 76. x =
0.16 and % fit = 2.2.
(v) Chemical shift in units of ppm. Data from Reference 77. x = 0.29
and % fit 4.1.
(w) Transition energy in kcal mole-'. Data from Reference 78. x =
0.06 and % fit = 3.6; x = 0.11 and % fit = 6.6.
(x) Dimensionless reactivity scale. Data from Reference 79. x = 0.02
and % fit = 7.8.
(y) Transition energy in kK. Data from Reference 80. x = 0.14 and %
fit = 6.2.
(z) Transition energy in kK. Data from Reference 80. x = 0.07 and %
fit = 6.7.
(aa) Asterisk indicates that limited data is available. The n-value is
doubled to take this into account.