EXTENSIONS OF THE ELECTROSTATIC-COVALENT AND UNIFIED
SOLVATION MODELS TO INCLUDE PHOSPHINE BASICITY AND HYDROGEN
BONDING SOLVENT POLARITY AND ACIDITY
STEVEN DOUGLAS JOERG
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
There are many people that have contributed to my success during my stay at the
University of Florida. I would like to thank each of them personally, but, in many cases,
this is simply not possible. Therefore, this short acknowledgment will have to suffice.
Graduate studies at the University of Florida would not have been the same
without my advisor, Dr. Russell Drago ("Doc"). He ran his research group in a manner
unlike any other. His "figure it out for yourself' approach enriched my knowledge and
ultimately will do the same for my career. Dr. Drago gave me so many opportunities to
learn from his experience and his example. I am saddened by the fact that, because of his
untimely death, he will not be able to see the results of all the effort he put into my
education. Dr. Drago is truly missed by me and by everyone else that has come into
contact with him over the years.
I must, of course, also thank Doc's wife, Ruth. She and Doc regularly opened
their homes up to the group for summer and Christmas parties. The time I spent at the
beach provided a nice break from research (even if all that Doc wanted to talk about was
my work). Ruth is an excellent cook, and I always looked forward to seeing what she
would put together. She has been a large part of many graduate student's lives, and she is
Many people have worked in the office as Doc's secretaries over the years that
I've been here. Most notable are Maribel Lisk and Diana Williamson. They each have
made my work easier with their efficiency and attention to detail. Diana must be thanked
for her non-stop work on my second publication. During my first summer, Maribel told
me that whenever I needed something done, that I should come to her. Now invariably
whenever I talk to Maribel, I preface my statement with "Maribel, you said I should come
to you whenever I had a problem." She has always been up to the challenge of my
numerous needs. I must also thank the many student assistants that have been a part of
the successful operations in the office.
In the introduction to his Reflections on the Psalms, C.S. Lewis wrote: "It often
happens that two schoolboys can solve difficulties in their work for one another better
than the master can . The fellow-pupil can help more than the master because he knows
less. The difficulty we want him to explain is one he has recently met. The expert met it
so long ago that he has forgotten." This has been my experience so many times
throughout graduate school. Students help students in their work, and then we turn to the
expert for final confirmation. We learn by solving our own problems. Many have helped
me to do this during my time here.
During my first summer, Chris Chronister, Melissa Hirsch, and Don Ferris
introduced me to work in the Drago group and to the area of E and C in general. Chris
and Don have continued to be resources to which I have turned. Don has always
graciously dug into his old research files in search of bits of information that might be of
use to me.
Many postdocs have come and gone during my stay. These include David Singh,
Phil Kaufmnann, Garth Dahlen, Doug Bums, and KrzysztofJurczyk. Many visiting
scientists have also helped me including Mariana Torrealba, Joaquim Sales, and Jim
George. Many thanks are given to Mariana for her help with chemistry and with Spanish
and for the hospitality of her and her husband.
Graduate students have been the true source of help and inspiration while I have
been here. Those that came into the group the same year that I did are Nick Kob, Alfredo
Mateus, and Churchill Grimes. Nick always tried to help me understand heterogeneous
catalysis, at least when he was not tanning himself on a University Auditorium park bench.
Alfredo came in the first summer with me and taught 2046L that first year. He has been a
good friend, computer guru, crossword puzzle co-worker, and fellow voyageur to Burrito
Other Drago students that have been a part of my life here include Karen Frank,
Mike McGilvray, Scott Kassel, Jose Dias, Silvia Dias, Mike Robbins, Ken Lo, Mike
Gonzalez, Andy Dadmun, Ed Webster, John Osegovic, Ben Gordon, Cheng Xu, Brian
Scott, and Andrew Cottone. I have enjoyed getting to know Karen during her second stay
here. Job hunting together has been fun (once they started to call). Mike McGilvray was
always one that I could turn to when I had a mechanical need. He always seemed to be
able to fix something that I broke or to otherwise lend a hand. Jose and Silvia must be
thanked for always having a kind word and a smile. I won't forget either those occasional
(usually chocolate) snacks that Silvia had for me. To the West Florida crew, Ed and John,
I must thank them for helping me to dig deeper in my understanding of chemistry and for
many helpful discussions of my work that helped to make it better. I know, however, that
we all did enjoy the day when John chose not to talk in group meeting. Doc enjoyed it
Other graduate students in the department have also made my stay enjoyable.
These include Candace Seip, Houston Byrd, Jon Penney, Mary Schmidt, Jennifer Wild,
Sid Parrish, and Jim Murphy. Brian Scott, Mary, and Candace all helped me through that
one particular semester of 2046. They must be thanked for making a difficult situation go
very smoothly. Candace especially helped during the lectures in trying to find something
we could do to be entertained. Our staff meetings that semester were truly interesting
(What solid?). Jennifer, Sid, and Jim all helped through other semesters of teaching just to
make the lectures and student problems bearable, if not enjoyable. Candace and Houston
both have been beacons of spiritual support during my stay. Conversations with them and
prayers from them have encouraged me in my Christian walk in ways that they cannot
Many professors have helped me in my decision to pursue teaching as a career.
Pam Clark, Bruce Weber, William Birdsall, Karen Campbell, Judy Geiser, and Rod Webb
have played that role to rave reviews. My small college education gave me the
opportunity to know them very well. Although I only had one lab class with her, Karen
Campbell always had the time to spend talking with me about school throughout my stay
at Albright. In graduate school, Gardiner Myers and James Horvath have helped me to
improve my teaching and have taught me to demand excellence from my students.
Two "little" guys must be thanked immensely for their help to me, even though
they will never understand what they have done. These two are Robbie and Andrew, my
"brothers." They have always given me something to look forward to on the weekend,
especially when research or teaching seemed like it would never end. I hope that I have
been a light in their lives as much as they have been in mine.
My parents must also be thanked for their sacrificial giving to my life. They have
always urged me to reach as far I could with my life. They want what is best for me, and I
appreciate that so much. Their love and support in so many areas have been invaluable.
TABLE OF CONTENTS
ACKNOWLEDGMENTS ........................................... ii
ABSTRACT ...................................................... ix
1 INTRODUCTION ...............................................1
Single Parameter M odels .......................................... 1
Dual Parameter M odels ........................................... 3
One Parameter Plots in a Two-Parameter World ........................ 14
M multiple Parameter Approaches ..................................... 18
Solvatochromism ................................................ 22
Models Used to Quantify Solvent Effects ............................. 24
2 REACTIVITY OF PHOSPHORUS DONORS .......................... 31
Introduction .................................................... 31
Calculations ......................... ....... .................. 37
Results and Discussion ............................................ 46
Conclusions .................................................... 71
3 EXTENSION OF THE ELECTROSTATIC-COVALENT MODEL TO
2:1 ADDUCTS ................................................ 74
Introduction .................................................... 74
Calculations .................................................... 78
Results and D discussion ............................................ 80
Conclusions .................................................... 104
4 DONOR-ACCEPTOR AND POLARITY PARAMETERS FOR HYDROGEN
BONDING SOLVENTS ........................................ 106
Introduction ................................................ ... 106
Experimental and Calculations ...................................... 108
Results and D discussion ............................................ 112
Conclusions .................................................... 135
5 MESITYL OXIDE AS A PROBE OF SOLVENT POLARITY AND ACIDITY. 136
Introduction .................................................... 136
Experimental ................................................... 138
Calculations ..... .............................................. 138
Results and Discussion ............................................ 144
Conclusions .................................................... 150
REFERENCES ................................................... 151
BIOGRAPHICAL SKETCH ......................................... 163
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
EXTENSIONS OF THE ELECTROSTATIC-COVALENT AND UNIFIED
SOLVATION MODELS TO INCLUDE PHOSPHINE BASICITY AND HYDROGEN
BONDING SOLVENT POLARITY AND ACIDITY
Steven Douglas Joerg
Chairman: Dr. David E. Richardson
Major Department: Chemistry
The understanding of how molecules interact is a fundamental step toward creating
more efficient chemical processes. Many models have been developed that characterize
molecules as donors and acceptors. These models usually calculate empirical parameters
for donors and acceptors based on their interaction with each other. If a particular model
is effective, using these parameters, one can then make predictions about how a given
donor and acceptor will interact without actually running the experiment.
One highly successful means of correlating donor-acceptor chemistry is the
Electrostatic-Covalent (ECW) model. This particular model assigns parameters to each
donor and acceptor based on the degree to which the molecule interacts electrostatically
and covalently. This research has brought a very important class of donors, the
phosphines, into the ECW model. Because phosphines are widely used as ligands in
organometallic chemistry, prediction of their reactivities is extremely important. This
research also examines the role that steric effects play in phosphine reactivity. Different
attempts to correlate the chemistry of phosphines are examined in detail.
Reactions are often affected by the solvent in which they are run. Very often the
solvent itself is a donor or acceptor, and a competition may occur between the reaction of
interest and the solvent. The Electrostatic-Covalent model correlates data only in poorly
solvating solvents, and one extension (the Unified Solvation Model, USM) has included
the addition of a term to take into account a solvent that interacts. The data used to
create many other scales of solvent polarity are all correlated with the USM, thus making
only one scale necessary. This research correlates the reactivity of many probe solute
molecules in solvents, including those that hydrogen bond. The USM is able to separate
specific and non-specific contributions to reactivity, something not possible with other
models. The reactivity of one particular solute, mesityl oxide, is studied in detail because
it has the potential for use as a probe of solid acid reactivity.
Many methods have been proposed in the literature'9 for the correlation of acid-
base and solvation chemistry. Each method has its basis in a set of fundamental
assumptions. Some of the more common approaches will be discussed below with
attention paid to those assumptions. Criticisms are outlined for each of the models
presented, and conclusions are drawn about the effectiveness of each. This dissertation is
concerned with extensions of the Electrostatic-Covalent Model (ECW) of donor-acceptor
chemistry and the Unified Solvation Model (USM).
Single Parameter Models
There have been many attempts in the literature to create scales of acidity or
basicity based on a single physical property or measurement with one particular acid or
base. For the most part, these have been successful for a specific case. These methods
assume that there is an inherent scale of acidity or basicity. (For example, a given base is
always stronger than another given base toward all acids.) This is a poor assumption, and
a graphical method for showing this will be presented.
A few of the more common one parameter examples will be presented here, and
then later in this introduction, reasons will be given for their limited success. These one
parameter plots are being done in a two parameter world, and under certain conditions,
good results are found. However, their use is not recommended because many
assumptions are made and the presence of two different contributions to a bonding is
Gutmann's scale' of Donor Numbers (DN) is a classic example of a one-parameter
plot. It is defined as the enthalpy of reaction of a dilute solutions of base and SbCl5 in 1,2-
dichloroethane. With this scale established, one can plot the enthalpy of interaction of a
base with any other acid versus the Donor Number. The results are usually linear. Several
assumptions5 are made with this method. The major assumption is that the scale of acidity
with any acid will be the same as that established with SbCI5, that is, that there is one
inherent scale of acidity (the one found with SbCl5). Despite this faulty assumption, the
DN scale has met with great success9 in areas such as predicting NMR chemical shifts,
variation in 0-H bond length through infrared measurements, ligand substitution reaction
rates and a wide range of other acid-base enthalpies.
Mayer, Gutmann, and Gerger2 have also introduced a scale of acidity called
Acceptor Numbers (AN). This scale is based on the 31"P chemical shift of(C2H5)3PO
(TEPO) when it forms an adduct with a given acid. (C2H5)3PO has also been used as a
probe of specific and non-specific interactions with various solvents. TEPO is a good
reference base for many reasons. The ethyl groups are large enough to shield any
alternate coordination sites, but too small to sterically hinder coordination of acidic
solvents.9 As with donor numbers, the assumption is made that the scale created with this
reference base will be the same regardless of the base studied. This scale will be discussed
in Chapter 4.
Many other one-parameter scales are seen in the literature. One of the most
common is to plot an enthalpy or spectral shift versus the pKA of a series of acids or bases.
One other scale that has been proposed,3 similar to the Donor Numbers, is the enthalpy for
coordination of a base to BF3 in methylene chloride solution. Although these one-
parameter plots are widely used (and frequently work well), they are limited by the
assumption that inherent scales of acidity and basicity exist. Several two-parameter scales
have been developed to take into account the different aspects of ar bonding.
Dual Parameter Models
The Edwards Equation
Edwards began his study of two parameter models with the correlation of
equilibrium constant and rate data for nucleophilic displacement reactions.4 He correlated
the data to Equations (1-1) and (1-2), where En is a nucleophilic constant characteristic of
log (Kn/Kref) or log (kn/krf) = En + 1 Hn (1-1)
log (Kn/Kf) or log (knkrf) = ac Pn + (3 Hn (1-2)
an electron donor, Hn is the basicity of the donor relative to the proton, P. is the
polarizability of the donor [calculated as log (RBn/Rwa.t), where R is the refractive index],
and a and 3 are substrate constants. The reference used is water (which is assigned E, = 0
and Hn = 0 values). The H, values are defined as 1.74 + pK, of base B,, and the En is
defined as 2.6 + Eox for base B.. Edwards and Pearson4 concluded that nucleophiles and
substrates tended to sort into two categories: those whose displacement reactions
correlated well with the pK, of the nucleophile and those correlating well with F, or P. of
About this same time, Ahrland, Chatt, and Davies6 recognized that metal ions
could be divided into classes depending upon whether they formed their most stable
complexes with the first ligand atom of each group (class a) or whether they formed their
most stable complexes with the second or subsequent member of each group (class b).
Class (a) metal ions include alkali metal, alkaline earth metals, and lighter transition metals
in high oxidation states. Class (b) metal ions are mostly heavier transition metals in low
Hard/Soft Acid/Base Theory (HSAB)
Pearson8 recognized consistency in this classification system and began to classify
potential acceptors as "hard" [class (a)] and "soft" [class (b)]. He expanded these
designations to include a wide variety of Lewis acids. He proposed that donors also have
this quality of "softness" to varying degrees. Hardness and softness are related to the
polarizability of the acid or base. Thus, those acids or bases with small size, high positive
charge, and no easily distorted or removed electrons are called "hard." Those acids or
bases with large size, small positive charge, and easily distorted electron clouds are termed
"soft."9 The general rule is that hard acids tend to prefer hard bases, and soft acids prefer
soft bases. This is observed from experimental evidence. Pearson refers to the hard-soft
principle as ". . a simple, but imprecise, law with a wide range of applicability.''8c Many
physical properties and reactivity patterns have been explained using hard/soft principles.
These include geometry trends, organic nucleophilic-electrophilic reactivity, selection of
organic catalysts, and solubility rules for inorganic salts. References to these and many
other chemical phenomena using hard/soft explanations are given in reference 5.
As one example, the solubility of salts is predicted fairly well with hard/soft
principles. Generally speaking, insoluble salts will be those that are combinations of ions
that are hard-hard or soft-soft. For example, AgI, a soft cation (acid) combined with a
soft anion (base), is very insoluble (KYp = 9 x 10.17). The principle is not absolute,
however. The chloride ion is classified as a hard base, thus making the prediction that
AgCl would be soluble. It is, in fact, insoluble. Also it is observed experimentally that
AgCl is soluble in an ammonia solution forming Ag(NH3)2+. Thus, the soft silver ion gives
up a borderline to hard base (CI) in favor of a hard base (NH3). Another example is
Ag2SO4, the combination of a soft cation with a hard anion. This should be a soluble salt,
but it observed to be only slightly soluble (Ksp = 1 x 10"5).
Exceptions to the hard/soft rules are observed when one does not consider the
relative strength of the acids and bases. Pearson began his studies looking at one-
parameter scales of acidity and basicity, but he knew "that reactivity of a Lewis acid or
base depends] on something more than a single monotonic scale of relative strengths" (p.
17).5 Here he began a more systematic study of the phenomena of"softness." Following
a suggestion by D.H. Busch, Pearson called it the "hard/soft parameter," with soft
corresponding to a large value of the parameter and hard to a small value.5 This
nomenclature is imprecise, simply because these words are normally used as opposites, but
in this case, they refer to the same property, but to different degrees.
Pearson introduced Equation (1-3) to show the contributions of strength (SA and
SB) and softness (aACTB) to the overall equilibrium constant.
log K = SASB + CaAaB (1-3)
This equation is not meant to be used quantitatively, but it is given merely to show the
different contributions. The hard-hard and soft-soft rules work well when hard species are
also strong and when soft species are weak.5 Equation (1-3) is very similar to the
equations proposed by Edwards. Yingst and McDaniel"0 observed that acids classified as
"hard" in general have low ax values, meaning they are sensitive to nucleophilic ligand
character or polarizability, and high 3 values. Acids classified as "soft" tend to have the
opposite (larger ac, smaller (3) values. They propose the use of the absolute value of (o/3)
to assign "hard" or "soft" designations. Pearson8c equates the 03Hn term with his SASB
term and cEn term with his CTAB term, but admits that Edwards' equation is not the only
one that might be related to Equation (1-3).
Molecular orbital theory can be applied to hard and soft acids and bases. Soft
species are characterized by a high density of low-lying empty orbitals and/or by a high-
lying HOMO. Hard species are characterized by isolated ground states and/or by a low-
lying HOMO.5 The base HOMO and acid LUMO are closer in energy in a soft-soft
interaction, and the stability gained from the HOMO-LUMO interaction brings the
bonding molecular orbital to a lower energy than the acid HOMO. This is characteristic of
a covalent (electron-sharing) interaction. In the hard-hard interaction, the acid LUMO
and base HOMO are more spread apart in energy, and the stabilization gained brings the
bonding molecular orbital to a higher energy than the acid HOMO. This is a more
electrostatic interaction, and the acid and base must react by means of any net charges or
permanent dipoles they happen to possess.5
Criticisms of HSAB
Many criticisms of hard-soft acid-base theory have been published. Myers"1
questions Pearson's use of polarizability in defining hardness and softness. In general,
Pearson describes something as "soft" if it is polarizable and "hard" if it is not. He has
been careful to point out that that polarizability is not the only factor which influences
softness. Myers gives many examples of metal ions that are supposed to be hard that
actually have larger polarizabilities than ions that are soft. He believes the terms "hard"
and "soft" are inadequate and that the relationship between class (a) and class (b) metal
ions needs to be re-examined to find an effect other than polarizability which distinguishes
Perhaps the most vocal criticisms of HSAB have come from Drago. 12 His primary
criticism is that HSAB defines hardness and softness as opposite properties, meaning that
something that is hard cannot be soft. This is illustrated by the binding of(CH3)3N,
(CH3)20, and (CH3)2S to phenol (hard acid) and 12 (soft acid).12' Toward phenol, the
enthalpies would be expected to follow this trend: (CH3)20 > (CH3)3N > (CH3)2S (hardest
base binding with the highest enthalpy). Toward I2 the opposite trend is expected (softest
base binding with the highest enthalpy). However, in measuring these enthalpies, only
four of the six are correctly ordered. Toward 12 the nitrogen donor is "softer" than the
sulfur donor, and toward phenol, the nitrogen donor is "harder" than the oxygen donor.
Clearly, something cannot be both soft and hard at the same time.12,
Purcell and Kotz13 have analyzed around 37,000 donor-acceptor pairs, trying to
predict which ones would form stable adducts. They based their prediction on the
similarity in hardness or softness (using Drago's C/E ratio, vide infra) of the donors and
acceptors. They found that they made the correct prediction only 40% of the time. It is
surprising that HSAB seems to work so well given these findings.
The Electrostatic-Covalent Model (E and C)
In 1965, Drago and Wayland14 proposed a four-parameter equation [Equation (1-
4)] for predicting enthalpies of adduct formation in the gas phase or in poorly solvating
-AH = EAEB + CACB (1-4)
solvents (e.g., CC14 or cyclohexane). In the equation, EA and EB are defined as the
susceptibility of the acid and base, respectively, to undergo electrostatic bonding. CA and
CB are the susceptibility of the acid and base, respectively, to undergo covalent bonding.
Mulliken's work15 on the covalent-electrostatic nature of the adduct bond is the
theoretical basis for the Electrostatic-Covalent model. In brief, Equation (1-5) gives the
WG(A,B) = avo(AB) + bi(A-B') (1-5)
Mulliken charge-transfer or valence bond model for a donor-acceptor interaction, where
the total ground-state wave function for the complex AB is given by WG. The wave
function W0o describes the interactions in the complex in which classical intermolecular
forces such as ion-dipole, dipole-dipole, and London dispersion forces are involved. The
function V,, arises from the overlap of the filled base orbital with the empty acceptor
orbital. 5 The coefficients a and b indicate the relative importance of electrostatic and
Drago and Wayland's initial method14 of determining electrostatic and covalent
parameters for acids and bases differs greatly from the method currently employed videe
infra). They arbitrarily assigned EA=I .00 and CA=I.0O to the acid 12. The parameters for
four amines [NH3, CH3NH2, (CH3)2NH, and (CH3)3N)] were the first to be determined.
They postulated that the EB for each base should be proportional to the molecule's dipole
moment (4), and the CB for each base is proportional to the total distortion polarization of
the molecule (RB). Using four simultaneous equations in the form of Equation (1-6), they
AH = at + bRB (1-6)
estimated the relative contributions of electrostatic and covalent bonding [coefficients a
and b from Equation (1-5)] to the enthalpy ofadduct formation. Because EA and CA were
assigned to be 1.00 for 12, the sum of the EB and CB parameters (EB = ag and CB = bRa)
will be the adduct formation [see Equation (1-4)].
After these amine donor parameters were determined, more typical methods of
determining acid parameters were then used. Several simultaneous equations in the form
of Equation (1-4) could be written, and keeping the EB and CB parameters constant, one
can calculate EA and CA for any acid for which enthalpies of adduct formation are known.
Currently, master-fit programs are used, with which one can enter enthalpies for hundreds
of adduct formations. The program will then determine the best acceptor and donor
parameters that will fit the enthalpies. Further information on this program will be
included in Chapter 2, where it is used to study phosphine reactivity.
The E and C equation underwent a minor change with the work of Guidry and
Drago. 16a They used previously measured enthalpies of adduct formation between
1,1,1,3,3,3-hexafluoro-2-propanol, HFIP, and a series of Lewis bases. When trying to
correlate the enthalpies to Equation (1-4) (fixing known EB and CB parameters for the
donors), a poor fit resulted. They attributed the source of the problem to intramolecular
hydrogen bonding between the hydroxyl proton and fluorines on the -CF3 groups. It was
determined videe infra) that upon adduct formation, this intramolecular hydrogen bond
was broken, at the cost of approximately 1.1 kcal mol'. This constant contribution to the
measured enthalpy does not depend upon the base added and must be expended before
adduct formation can take place. The measured enthalpies were then 1.1 kcal mol"' less
exothermic because this amount was needed to break the hydrogen bond.
Guidry and Drago proposed an extension to the E and C equation similar to
Equation (1-7), where W is any constant contribution to an enthalpy that is not associated
AH = EAEB + CACB + W (1-7)
with the specific donor-acceptor interaction. As in this case, some change needed to be
made to the acceptor before it could react with the donor. Other examples of this type of
constant contribution would include breaking a dimer, in order for the monomers to react
as acceptors. 16b An endothermic W is entered as a negative number into Equation (1-7),
to decrease the magnitude of the calculated -AH.12a
Using Equation (1-7), the enthalpies of adduct formation, and the known EB and
CB parameters for the donors, Guidry and Drago solved a series of simultaneous equations
for the best EA, CA, and W to fit the data. The W value was 1.1 kcal mol-', the energy
needed to break the hydrogen bond. Thus, Equation (1-7) provides an indirect method for
calculating the energy needed to break the hydrogen bond or, in other cases, a dimer.16b
In the case of a dimer, the W value is actually one-half the enthalpy of dimerization
because all -AH values are for formation of a mole of adduct (which uses one monomer,
not two). These enthalpies cannot be obtained by other methods, so this calculation
becomes very valuable.
An alternate form of Equation (1-7) may also be used to correlate any
physicochemical property such as spectral shifts (e.g., UV/Vis, NMR, or IR), reaction
rates, and equilibrium constants. This form is given in Equation (1-8), where AX is the
AX = EA*EB+CA*CB + W (1-8)
value of the physicochemical property (with units of energy), and W is the value of AX
when EB=CB=O. The asterisks on the acceptor parameters indicate that changes in AX are
being measured with a constant acceptor and a series of donors. The asterisks also
indicate that these parameters are not based on enthalpies and have different units
incorporating conversion units to give AX in the units of the measured property. When
correlating reaction rates or equilibrium constants, one must take the log of the value in
order to meet the energy unit requirement.
The most current set of donor and acceptor parameters17 was created by fixing the
iodine EA and CA parameters at 0.50 and 2.00, respectively. Also fixed were the EB
parameter for N,N-dimethylacetamide (DMA) at 2.35 and the CB parameter for (C2H5)2S
at 3.92. Over 500 enthalpies ofadduct formation for reactions of about 50 bases and 43
acids were solved simultaneously to give a unique solution. Fixing the parameters
described above at different values would bring about another unique solution.
Many insights can be gained through an ECW analysis of a data set. The model is
designed to predict the normal a bonding energy for adducts. When a particular system
does not fit the model, it is because factors other than c bonding are operative. These
deviations should be used to design new experiments to help elucidate all factors that
contribute to an enthalpy or spectral shift.
Several specific effects can be observed through ECW analysis. The most
common effect that is observed is an enthalpic steric effect. This effect can be assigned
when the enthalpy predicted by the model is more exothermic than the experimentally
measured enthalpy. The steric bulkiness in the donor-acceptor interaction does not allow
optimal orbital overlap. The difference between the experimentally measured and
predicted enthalpies is the strain energy, a value that cannot be directly measured.
The existence of it-back-bond stabilization can also be observed with the ECW
analysis. This effect is assigned when the predicted enthalpy is smaller than the
experimental enthalpy. Because ECW only models a interactions, the predicted enthalpy
will be smaller. Experimentally both a and nt bonds are being formed. Chapters 2 and 3
will present a model showing how these 7t effects can be incorporated into the a-only
acceptor parameters, with good results. Unfortunately, when this occurs, the parameters
are difficult to interpret because the contributions from the two types of bonding cannot
Criticisms of the ECW Approach
The early criticisms of the ECW approach have been treated with extensions to the
original form of the model. The more noteworthy extensions resulted because
experimenters wanted to treat systems with strongly solvating solvents.5 Equations (1-4)
and (1-7) were developed using data in poorly solvating solvents, such as cyclohexane. A
later section of this introduction will deal with an extension to Equation (1-7) to include
wide range of solvent polarity. This extension is termed the Unified Solvation Model
Other areas that were criticized for not correlating to the ECW approach include
systems having large amounts of charge transfer or those having ions as reactants instead
of only discrete, neutral species.9 Another extension of the ECW Model, Equation (1-9),
AH = EAEB + CACB + RATB (1-9)
has been proposed for just these types of reactivity. In Equation (1-9), RA is used to
indicate that the acid is the acceptor in the electron transfer interaction, and TB indicated
that the base is the transmitter. These parameters are available for use,' and another
correlation of homolytic bond energies (using terminology such as "catimers" and
"animers") should also be examined2''28 for further insight into the use of Equation (1-9).
One general criticism that is discussed5'13 relates to the arbitrary nature of fixing
four parameters initially. While this is needed in order to find a unique solution of the
parameters for all acids and bases involved, deciding which parameters to fix initially is
arbitrary, and changing these will give a different set of parameters. Jensen writes, "... .
its use of purely empirical numbers disguises any relationship between Lewis acid-base
reactivity and chemical periodicity" (p.20).5 While this is a drawback, ECW appears to be
able to correlate a wide range of donor-acceptor interactions. As described above, its
extensions help to fill in areas that the original equation was unable to correlate, thus
increasing its usefulness.
One-Parameter Plots in a Two-Parameter World
As discussed above, one of the major assumptions that is made with one-parameter
plots is that there is an inherent scale of acidity or basicity. This means, for instance, that
one particular acid is always a better acid than another regardless of the base. Drago's E
and C approach shows that it is possible for a given acid to interact better electrostatically,
but not covalently, than another acid. This allows for a varying scale of acidity or basicity.
Cramer and Bopp20 have shown how the scales can change depending upon which acid
and base are interacting. (The reader should note that Cramer and Bopp's article was
written before the most current set of E and C parameters were calculated. Present use of
their model requires the use of the new parameters12'). They present a graphical
representation using Equation (1-10).
-AH CA + EA (CB + EB) (CA EA)
= + (1-10)
(CA + EA) 2 2 (CA + EA)
Figure 1-1 is an example of the type of graph that can be created. The x-axis is the
quantity (CA EA) / (CA + EA), thus showing acids moving from less covalent in nature to
more covalent (as one goes from negative x to positive x). The y-axis represents -AH /
(CA -EA). Therefore, if one knows the EA and CA parameters for a particular acid, the
enthalpy of interaction can be calculated from the value on the y-axis.
It should be noted that the slopes of the line presented in Figure 1-1 are based on
the enthalpies of interaction of the donors with two acceptors, phenol and iodine. Other
slopes varying slightly could be obtained using the enthalpies of interaction with other
acids. This graph indeed shows the danger of assuming that one donor is always stronger
than another. When one asks which of two donors is stronger, the next question should
be when interacting with which acceptor? The degree of covalency is extremely important
when predicting enthalpies of interaction, and this is largely ignored in the literature when
one-parameter plots are made.
more electrostatic < > more covalent
(CA- EA)/(CA+ EA)
... pyridine ammonia ------- DMSO
.......acetonitrile -- diethyl sulfide
Figure 1-1. Five donors interacting with acids of varying CA/EA ratios
Dragol2"'9a has outlined the reasons that one-parameter plots often appear to give
good correlations. A modified version of ECW is given [Equation (1-11)] by
= CB + EB (1-11)
dividing both sides by EA. If the entire right side of Equation (1-11) is called Bi, where i
represents the CA/EA ratio of the acceptor, Equations (1-12) and (1-13) result.
-AH = EABi (1-12)
Bi = EB + iCB (1-13)
If, for example, one uses i = 0.2 (i.e., some acceptor property has a CA/EA value of 0.2),
an order of donor strength can be calculated using Equation (1-13) and the EB and CB
values of the donors involved. Table 1-1 shows Drago's calculated'2' B0.6 and Bo.2 values
for several donors along with their EB and CB values. The examples of CA/EA ratios of 0.6
and 0.2 are given here because they represent very common ratios for physicochemical
properties. This table is another way of viewing Figure 1-1. Notice the difference in
donor strength order that is found depending upon the CA/EA ratio of the acceptor studied.
Successful one parameter correlations can be made when one plots their
physicochemical property (e.g., a measured enthalpy or chemical shift) versus B0.6, Bo0.2, or
any similarly prepared scale if the new property has the same CA/EA ratio as that used to
make the scale. The most common example of this is seen when plots of experimental
data are made versus pKA or pKB. Drago reports'2' that the CA/EA ratio for pKB values is
0.2. Therefore, any physicochemical property that also has a CA/EA ratio of 0.2 will plot
linearly with pKB. As stated earlier, 0.2 and 0.6 are very common CA/EA ratios, and linear
plots will often result.
It would also be possible to find a nearly linear plot to pKB with experimental data
that have a CA/EA ratio of 0.6. Many of the points will fall on the best fit line, and a few
may deviate. Explanations are given for the misses that are encountered with these points.
Table 1-1. EB, CB, Bo.6, and Bo.2 values for several donors
Donor [ EB CB Bo.6 Bo.2
C5H5N 1.78 3.54 3.90 2.49
CH3C(O)N(CH3)2 2.35 1.31 3.14 2.61
(CH2)40 1.64 2.18 2.95 2.08
(C2H5)20 1.80 1.63 2.89 2.22
(C2Hs)2S 0.24 3.92 2.59 1.02
CH3C(O)OCH3 1.63 0.95 2.50 1.99
CH3CN 1.64 0.71 2.07 1.78
Drago believes that these explanations are often incorrect and do not credit the amount of
covalency involved in the interaction. He also warns about the dangers of not using a
series of very different donors (with differing CB/EB ratios) in an analysis. Without a wide
range of donors, potentially outlying points may be not be observed, and a good
correlation could be determined incorrectly. Unfortunately, a majority of experimental
data found in the literature use donors (or acceptors) from within a single family (e.g., all
pyridines or all phenols). Drago has written2' about the limitations of such analyses.
Multiple Parameter Approaches
Giering and coworkers22 have developed a multiple parameter equation for the
correlation of physicochemical properties. It takes the general form of Equation (1-14),
property = ax + bO + c(O 9g)k + dE,r + e (1-14)
where X is an electronic parameter that describes the intrinsic electron donor capacity of a
ligand videe infra); 0 is a steric parameter, the Tolman23 cone angle videe infra); (0 0st)k
is the "steric threshold" term, with Ost the cone angle above which abrupt changes in steric
effects occur; X is a switching factor, either 0 or 1 depending on whether or not the
particular ligand has a cone angle larger than 0,t for that system; and Ear is the "aryl effect"
parameter videe infra). This equation (known as the Quantitative Analysis of Ligand
Effects or QALE) has been used successfully to correlate thermodynamic, kinetic,
spectroscopic, and structural data of carbon, nitrogen, phosphorus, arsenic, and silicon
Early QALE studies24 included ligands with aryl, alkyl, and alkoxy substituents and
focused on the development of three classes of phosphorus donors, depending upon their
ability to act as a a-donor and 7t-donor (class I), only a o-donor (class II), or as a o-donor
and 7t-acceptor (class III). These classes have not been emphasized in the more recent
literature. Giering's attempts to calculate the relative contributions of a and 7t bonding
have been examined25 and found to be unsatisfactory.
Ligands with P-O (e.g., phosphites) and P-F bonds were studied in the early
literature together with alkyl and aryl-substituted ligands. These were believed to belong
to class I, ligands having both o- and 7t-donor abilities. Phosphites often missed in the
correlations, and these deviations have been attributed to the it-donating ability of the
ligands. Poor correlations resulted because of the x parameter, which is defined as the
difference between the A, terminal carbonyl band of LNi(CO)3 and 2056.1 cm1 [the A1
band for (t-Bu)3PNi(CO)3]. It was found that x must be modified for those donors that
could also have a it-donor contribution to bonding. A new parameter, Xd, based upon pKA
and Tolman's cone angle, is proposed26 that will allow better correlations. Despite this
new parameter, phosphites are still not studied with other ligands in the QALE analyses.
The Aryl Effect
Upon examination of several systems using only their X and 0 terms, Wilson and
coworkers found unsatisfactory results. They found that the inclusion of an additional
electronic parameter, Ear, improved the correlations.27 The Ear parameter values are
dependent only on the number of aryl groups present on a ligand and not on the
substitution of those aryl groups. The Ear parameter takes values of 0, 1, 2, and 2.7 for
ligands with no, one, two, and three aryl groups, respectively. These values were
determined using plots of Taft's polar substituent constant, ZCo*, versus X and 0.27 It was
observed that the phosphorus donors were separated into three parallel linear regions,
corresponding to no, one, or two aryl groups on the ligand. The spacings between the
linear regions were about the same, so these Ear values were assigned one unit apart. For
the triarylphosphines (all with 0=145), no parallel line could be found on the plot of F5*
versus 0. The Ear value was assigned to be 2.7 as this optimized the r2 values of the
systems being studied.
The Ear parameter indeed improves correlation coefficients. The reason behind this
improvement remains a mystery to those that use Ea.27,28 It has been suggested that"..
the aryl effect is a consequence of incipient nt-stacking interactions .. contributing to the
stabilisation of the transition state. These interactions would be dependent on the number
of aryl groups and independent of the nature and position of the substituents on the ring"
(p. 232).28b Further thoughts (from the perspective of ECW) on the nature of the aryl
effect will be given in Chapter 2.
There has been some concern in the literature about the use of the Tolman cone
angle in correlations such as these. Brown and Lee29 list several assumptions (also
discussed by Tolman2) that must be made if one is to use this parameter. One assumption
is the choice of the actual cone angle because the ligand's substituents can have varying
conformations. Tolman always chose the conformation that would give the smallest cone
angle. Depending upon the particular complex (i.e., considering other ligands on the
metal center) that is formed, this may or may not be the "actual" cone angle of that ligand.
Drago30 has discussed the need for rearrangement of an alkyl substituent chain in the
formation of the transition state. Conformational degrees of freedom may be lost during
this rearrangement, and he terms this an "entropic steric effect." H.C. Brown and
coworkers3 have also reported this effect.
Another possible difficulty with the use of cone angles is that". substituent
ligands bound to the same metal center can sometimes mesh with one another, permitting
closer packing of ligands than would be expected based on cone angle values" (p. 92).29
Brown and Lee write: "It is too much to expect that a single set of ligand parameters will
reflect accurately the relative steric requirements of ligands in markedly different situations
... Each ligand has not one, but a range of steric requirement values, depending on the
reaction or structural context" (p. 104).29 They also caution the use of parameters without
knowing their precision. White and Coville32 report that cone angles are good to within
2. Brown and Lee29 give many examples of the ranges of cone angles that have been
reported for phosphorus donors in various complexes.
Prior to the studies of Giering and coworkers, Tolman23 proposed the
"steric/electronic box," noting that both steric and electronic effects contribute to
reactivity. Giering's model is similar in form to the one used by Tolman, Equation (1-15).
z = aO + bv + c (1-15)
Here, 0 is the cone angle, v is the A, stretching frequency for Ni(CO)3L (where Giering
uses X as the difference in frequency from a standard), and z is any physicochemical
property. The coefficients a and b give the relative contributions from electronic and
Three excellent reviews2932'33 examine the new attempts to get a grasp on the
concept of cone angles and the steric requirements of ligands in general. Some of the
ideas include modeling solid angles, ligand radial profiles, and development of ligand
repulsive energy parameters, ER.
It is experimentally observed that many molecules have solvatochromic behavior,
meaning that their UV/Vis, IR, or NMR spectra are affected by the solvent used in the
measurement. These effects can be related to both specific and non-specific solvation.
Specific interactions are those that are related to acid-base interactions using specific
orbitals, including donor solute molecules in acceptor solvents and vice versa. Non-
specific effects cover a broader range including interactions of a solute dipole (permanent
or induced) with both the internal dielectric of the solvent cavity and the dielectric of the
bulk solvent. 12a
Solvent Effects on Electronic Spectra
An explanation often used12,'34 for changes in electronic spectra due to solvent
effects is based on the polarity of the ground and excited states of a probe molecule.
Figure 1-2 shows the changes in the transition energy of a molecule from its gas phase to
its introduction into a polar solvent. Figure 1-2a shows the case for a molecule whose
ground state is more polar than its excited state. In the polar solvent, the ground state is
Gas Phase Polar Solvent Gas Phase Polar Solvent
(a) Blue-Shifted (b) Red-Shifted
Figure 1-2. Changes in transition energies in moving from the gas phase to a polar
solvent. Two examples are possible: (a) the blue shift and (b) the red shift.
stabilized, and the excited state is destabilized, thus creating a larger ("blue-shifted")
transition energy. Figure 1-2b shows the case for a molecule whose excited state is more
polar than the ground state. Its excited state is stabilized in the polar solvent, creating a
smaller ("red-shifted") transition energy. Reichardt34b discusses the various combinations
of polar and nonpolar solutes and solvents in detail listing various minor contributions to
red and blue shifts.
This type of behavior is predicted by the Franck-Condon principle,34b1'3S which
states that because electronic transitions occur at a much faster rate than vibrations, the
nuclei of the chromophore are not changed during the electronic transition. Therefore,
when the excited state is formed, the solvent cage does not have a chance to rearrange. If
the molecule becomes more polar upon moving into the excited state, there is an overall
stabilization (in a polar solvent). If the molecule is less polar in the excited state, there will
be a destabilization because the solvent cannot rearrange during the electronic transition.
Solvent Effects on NMR spectra
NMR spectra are affected by solvents in two ways.34b Differences in the bulk
magnetic susceptibility (Xm) between solute and solvent is the first way. Normally the bulk
susceptibility differences are accounted for with the use of an internal standard, thereby
having the solute and standard each in a similarly shaped container. Another method that
is used to account for the bulk susceptibility differences is to take measurements as several
solute concentrations and extrapolate to infinite dilution, thus giving the chemical shift
under volume susceptibility conditions of the pure solvent.34a If an external standard is
used, corrections can be made using Equation (1-16), where 5exp is the experimentally
6cor = 6cxp + [(2n/3) AXm o 106] (1-16)
observed chemical shift, 5. is the corrected chemical shift, and AXm is the difference
between the magnetic susceptibilities of the reference and the solution.
The second way that chemical shifts are affected in the NMR experiment is
through interactions (both specific and non-specific) of the solute and solvent. Non-
specific interactions typically do not affect chemical shifts nearly as much as specific
interactions. Examples of the effect of solvent on NMR chemical shifts are given in
Chapters 4 and 5.
Models Used to Quantify Solvent Effects
Kirkwood Approach to Non-specific Solvation
Kirkwood36a and Onsager36b developed a model to calculate the solvation energy of
a solute in a given solvent. Their expression takes the form of Equation (1-17), where s is
the bulk solvent dielectric constant, 6, is the internal dielectric constant of the solvent
cavity, p. is the solute dipole moment, and b is the radius of the molecular cavity.
E.o = 2 [(e e)/(2e + e*)] (2/b2) (1-17)
There are difficulties in estimating the internal dielectric constant of the solvent cavity and
the radius of the molecular cavity. Without the use of approximations for these values, the
model is not especially practical quantitatively. 12a,37 For this reason, efforts have been
made to develop empirical models of specific and non-specific solvation using various
probe molecules.38' While some of the more popular models will be discussed in detail in
Chapter 4, a brief introduction to them will be given here. An excellent review38b tabulates
the parameters used in these various models.
Empirical Models Using Probe Molecules
One of the first scales that was developed was Kosower's Z-value.39 It is based on
the transition energy for the longest wavelength band observed in 1 -ethyl-4-
methoxycarbonylpyridinium iodide (pyridinium iodide 214) in a wide range of solvents.
Alternate derivatives of this probe are used for non-polar solvents. Griffiths and Pugh40
have expanded Kosower's original work to include 43 more solvents. They have
suggested some changes in the use of the Z-value.
Perhaps the most widely used scale of solvent polarity is based on the negative
solvatochromism of the pyridinium N-phenolate betaine dye 36 ofReichardt.34b,38a This
probe molecule's 7it-+ absorption band has been measured in over 350 solvents. This
probe is not soluble in some non-polar solvents, so an alternate probe, with t-butyl
substituents, is used. This probe is often used in the literature to correlate
Taft, Kamlet, and co-workers42 have developed and modified a scale that covers
both specific and non-specific contributions to solvation. Their model takes the form of
Equation (1-18), where XYZ is any physicochemical property, 7t* is a measure of solvent
XYZ = XYZo + s7rt* + aa + b13 (1-18)
dipolarity/polarizability, a is a measure of hydrogen bond donating ability, and 13 is a
measure of hydrogen bond accepting ability. Later versions of the model have included an
h(6H2) term using Hildebrand's solubility parameter43 as a measure of the contributions of
creating a cavity in the solvent to accommodate a solute, and a d5 term (with 6 here being
a polarizability correction term). Instead of using a single probe molecule, Taft and
Kamlet chose seven molecules, mostly substituted nitro-anilines and phenols, as the basis
for their scale. It was their original belief that using several probes would help to average
spectral anomalies that might be found in any single probe. Later analyses of the 7r* model
find that this averaging actually ". . blurs very important, physically meaningful
contributions" (p. 5808).42c These authors suggest using only 4-nitroanisole and N,N-
dimethylnitroaniline as the basis for a revised and extended 7t* model. As with ET(30), the
7t* model has been used extensively in the literature44 to correlate data. Drago45a has
expressed concern over the use of the 13P-7t* scale and has compared its use to ET(30) and
his Unified Solvation Model videe infra).
The Unified Solvation Model (USM)
Drago and coworkers45 have proposed an extension of the Electrostatic-Covalent
model of acid-base chemistry to include non-specific solvation. Equation (1-19) shows
AX = PS'+W (1-19)
the extension, where AX is any physicochemical property, S' is the non-specific solvent
parameter, P is the susceptibility of a solute probe molecule to solvation, and W is the
value of the physicochemical property when S' = 0. In the earliest work,45b 162 spectral
shifts (e.g., UV, IR, NMR) for thirty probes in thirty-one solvents were correlated with
Equation (1-19). A master-fitting program was used so that all unknown variables (P, S',
and W) could be allowed to float. Probes and solvents were chosen carefully so that only
non-specific interactions would be present between solute and solvent. S' values for
solvents and P and W values for probes were calculated.
Use of the term "probe" in this dissertation refers not only to the actual probe
molecule, but more often to a particular physicochemical property of that molecule. For
instance, if a molecule is chosen to be a "probe" of solvent polarity, it is actually a
particular property such as an electronic absorption or chemical shift that is the true probe
that is studied to gain the desired information. As stated above, probes have P and W
values. The sign of the P value gives important information. Probes with negative P
values exhibit red shifts in more polar solvents, and those with positive P values exhibit
Later additions to the USM45 increased the number of probes to forty-one and the
number of solvents to forty-six (using 366 spectral shifts). With the additional probes and
solvents, the P, S', and W parameters were recalculated. This work also examined
specific interactions with the use of hydrogen bonding solvents. Equation (1-20) gives the
AX = EA*EB + CA*CB + PS' + W (1-20)
combination of specific and non-specific interactions to AX, combining Equations (1-8)
and (1-19). Using the master-fitting program and the known P and W values for the probe
molecules, EA*, CA*, and S' were calculated for the hydrogen bonding solvents and EB
and CB were calculated for the probe molecules. Chapters 4 and 5 will present further
extensions to the USM using more probe molecules and hydrogen bonding solvents.
A graphical method is proposed45c for separation of specific and non-specific
contributions to an electronic transition. Drago and coworkers studied Reichardt's
betaine and ETr(30) scale 41'38, in various mixtures ofo-dichlorobenzene and hydrogen
bonding solvent. They began in pure o-dichlorobenzene and added small aliquots of
alcohol. After each addition, the electronic transition [ETr(30)] was measured in the
UV/Vis. For the first several additions, Er(30) increased rapidly due to the specific
interaction between betaine and the alcohol. After the majority of probe had been
specifically completed, the ET(30) values began to level out and not increase so rapidly,
now as a result of non-specific solvation. The plot of measured ET(30) value versus
concentration of alcohol added to the system resembles a type I adsorption isotherm (see
Figure 1-2). A straight line can be drawn through the flattened region at high alcohol
concentrations. This line can be extrapolated to ET(30) at zero alcohol concentration. The
difference between this value and the E-(30) value in the pure alcohol is labeled the non-
specific contribution of the alcohol to solvation. The difference between the first point of
0 ....---- .. ..---.... ---....---. ...
0 5 10 15 20 25
[alcohol], mol L'1
Figure 1-2 Graphical Method for Separation of Specific and Non-Specific
Contributions to Solvation
alcohol addition and this newly extrapolated point is the specific contribution of the
alcohol to solvation. The rest of the isotherm [from ET(30)=0 to the first point in pure o-
dichlorobenzene] is the non-specific contribution due to o-dichlorobenzene.
The Unified Solvation Cavity Model (USCM)
Recently introduced,46 USCM is an extension to the USM that includes a term for
formation of a cavity within the solvent to accommodate a solute molecule. Equation (1-
21) gives the USCM, where all variables have similar meanings as in Equation (1-20).
AX = QS"2 + PS'+ EA*EB + CA*CB + W (1-21)
Here Q is the solute contribution to cavity formation and S"2 is the square of the non-
specific solvent parameter, and represents the solvent contribution to cavity formation.
S'2 was chosen as the solvent contribution because of the work of Taft and others42.'f
discussed earlier, which uses Hildebrand's solubility parameter (6H2) in the cavity term.
Bustamante and Drago46 have shown a correlation between S' and 6H, and thus have
chosen S'2 as their solvent-dependent cavity term parameter. They find very good
correlations of solubilities and enthalpies of solution to Equation (1-21) using donor and
non-polar solvents with non-polar solutes. Bustamante and Drago found only a small
contribution from the E and C (specific) terms of Equation (1-21).
REACTIVITY OF PHOSPHORUS DONORS
The ECW model12a'21 ,30'47 was developed at about the same time that chemists
began to recognize the need for at least two independent effects to define basicity. These
effects have been called hard/soft, Class (a)/(b), charge/frontier, nucleophilicity/basicity,
and electrostatic/covalent. The ECW model, Equation (2-1), uses the enthalpies of
adduct formation measured in poorly solvating solvents to derive quantitative parameters
for a scale ofo a donor, B, basicity and a scale ofo a acceptor, A, acidity.
-AH = EAEB+CACB + W (2-1)
Two effects are needed to fit the data. The magnitudes of the parameters for these two
effects parallel quantitative hard-soft, etc., reactivity trends. The electrostatic, E, and
covalent, C, model was selected to name the parameters EA, EB, CA, and CB.
The enthalpy basis for ECW provides parameters that are related to donor-
acceptor bond strength and are free of the complications from solvation or entropic
effects. Omission of systems with nt-back-bonding contributions and steric effects
provides parameters free of these influences. In enthalpy analyses of new systems where
these effects exist, the parameters provide estimates of the magnitudes of steric strain and
Recently,3'4 a dual parameter substituent constant equation [Equation (2-2)] was
reported to analyze the reactivity of families of compounds whose E and C values are not
Axx = dEAEx + dCACx + AXH (2-2)
In Equation (2-2), Ax is the measured property for the molecule containing substituent
X, AXH, the value for the parent hydrogen compound, AEX and ACx give the proportional
change in the E and C values of the parent compound induced by the substituent, and dE
and dc gauge the sensitivity of the reaction to substituent change. The AEx and ACx
values are the dual parameter analogues of the Hammett o-values while dE and dc are the
dual parameter analogues of p. It is reported30 that the set of dual substituent
parameters correlates data that previously required different sets of substituent constants
The d-values of Equation (2-2) are related to the E and C values of Equation (2-1)
dAE = SBE EA* (2-3)
dAc = SBc CA* (2-4)
Equations (2-3) and (2-4) are written for an analysis in which a family of donors is studied
and the acceptor is held constant. The subscripts are changed when the donor is kept
constant and a family of acceptors studied. In Equations (2-3) and (2-4), the dAE of
Equation (2-2) and EA* of Equation (2-1) are related by the family dependent
proportionality constant sBE which measures the sensitivity of the E values of the family of
donors to the substituent change. The proportionality constant sec, which relates dAc to
CA*, measures the sensitivity of the C values of the family of donors to substituent change
relative to setting sBE = 1 and SBc = 1 for pyridine. Thus, an SB value > 1 indicates the
family has a greater basicity response to the substituent than pyridine, while SB < 1
indicates the substituent effect is transmitted less effectively.
It has been further shown30'48 that the EB or CB values of the X-substituted donor
EBX and CBx, are given by where EBH and CBH are the E and C values for the parent
EBX = SBE AEx + EBH (2-5)
CBx = SBC ACX + CBH (2-6)
The implications of Equations (2-5) and (2-6) are profound for they greatly expand the
number of donors that can be analyzed with Equation (2-1). For example, with sBE = 1
and SBc = 1 for substituted pyridines, EBX and CBX values for seventy-seven, 3- and 4-
mono substituted pyridines can be calculated with Equations (2-5) and (2-6) from
reported2 AEX and ACX values. In a similar fashion, with reported values ofsAE = -0.817
and SAc = -0.225 (see Chapter 4), the EAx and CAx values of seventy-seven 3- and 4-
substituted phenols can be calculated. Substituting the resulting Ex and Cx values for
pyridines and phenols into Equation (2-1) permits the calculation of 6,084 enthalpies of
interaction of various pyridines with various phenols. Calculations can also be made for
the reaction of all of these pyridines with all the acceptors whose EA and CA values are
reported and for the reactions of all these phenols with all the donors whose EB and CB are
It has been emphasized, that in using Equation (2-1) to solve for EA* and CA* to
characterize a new reaction, one should use donors with very different CR/ER ratios.
When this is not done, a shallow minimum exists in the data fit and the uncertainty in the
parameters exceed the errors determined from goodness of fit criteria. This is a very
significant problem in the interpretation of the dE and dc parameters of Equation (2-2). By
definition a substituent constant analysis restricts the study of a reaction to a single family
of donors (or acceptors). This restriction usually leads to a small variation in the CR/EB
ratio of the data set. Thus, if the EB and CB values are available, Equation (2-1) should be
used to characterize an acid property and instead of restricting measurements to a single
family, different donors with a wide range of CR/EB values should be studied. Increasing
the range provides more accurate values of EA* and CA* whose interpretation is more
reliable than that of dE and dc.
The reactivities ofphosphines have been analyzed30 with Equation (2-2) by
summing the AE and AC values of the substituents attached to phosphorus. In this
chapter, the ER and CB values oftrisubstituted phosphines will be determined directly.
This is particularly significant because of the importance of phosphines in organometallic
and catalytic chemistry. At present, only tentative EB and CB values are reported'2" for
two phosphines. Determination of Ea and CB will permit an evaluation of the validity of
the substituent summation used30 in the phosphine analyses and by providing more
accurate EA* and CA* values for several physicochemical properties lead to a more
meaningful interpretation of the influence of electronic, steric and nt-backbonding effects in
the reactions these compounds undergo with acceptors.
The EB and CB parameters provide a o basicity scale that can be used with
Equation (2-7) in correlations to determine if physicochemical measurements, X, are
controlled by the same factors that influence bond strengths. X can be a spectral shift, rate
constant, activation enthalpy, redox potential or any measurement that is expressed in
X = EA*EB +CA*CB + W (2-7)
Free energies can be interpreted with these enthalpy based parameters because the goal of
established linear free energy scales is to derive enthalpy related parameters in order to
interpret the correlation in terms of electronic effects. This is an important point that has
been overlooked in the organometallic literature. Parameter derivation must use
enthalpies or employ systems in which the enthalpies vary linearly with the entropy to be
meaningful. If the desire were to predict and understand entropies in non-linear free
energy systems, a separate entropy scale would be needed.
In analyzing x's for new systems where entropies do not vary in a linear manner
with enthalpies, frequencies are not linear with force constants, or NMR chemical shifts
contain neighbor anisotropic contributions, the correlations of these x's to ECW or
meaningful linear free energy parameters should fail. In failed correlations, the ECW or
linear free energy models have not failed, but rather X has contributions from effects not
related to bond strength. Thus, the model provides an understanding of reactivity and
spectroscopy even when poor correlations result.
The ECW model is unique in terms of its generality. No other set of reactivity
parameters encompasses the variety of donors and acceptors used in this model. This is a
very important point for any multiparameter correlation. In the study of new acceptors,
the EA and CA parameters used to correlate reactivity arise by solving simultaneous
equations of the form of Equation (2-7). These equations must be independent and show
a low correlation to each other in order to obtain a definitive fit. In acceptor correlations,
we have emphasized that the CB/EB ratio for the donors must vary in order to define the
EA and CA parameters accurately.'247' Unfortunately, in studies limited to phosphorus
donors the CB/EB ratios of the donors selected are often similar. When this is the case,
good data fits can result but the acceptor correlation parameters are without meaning.12'
Enthalpies cannot be predicted for any donor whose CB/EB ratio lies outside the CB/EB
range used and correlated systems may contain reactivity effects that are not related to
bond strength. The only reliable conclusion from the correlation of such a data set is to
spot a deviant system and to then proceed to investigate causes for its deviation.
Acceptor parameters from such analyses should be considered tentative.
There is widespread acceptance of the dual nature of donor-acceptor bond
strength in electrostatic/covalent, hard/soft or charge/frontier descriptions. With the
exception of the ECW model and early work by Edwards,4 literature correlation analyses
of reactivity have not employed parameters related to these quantities as measures of
substituent effects or basicity. Derivations are reported47b that show that each acceptor,
with a significantly different softness, requires a different one parameter basicity scale. As
a result, use of a one parameter scale to define basicity and substituent constant effects has
led to a proliferation of scales, each with limited utility.
This chapter reports phosphorus donor parameters which are well connected to the
other donors in ECW. A model is presented to support the idea that the R-acceptor
properties of the phosphines decrease regularly as their a basicity increases. It is shown
that the phosphine o donor, EB and CB, parameters inappropriately can correlate reactivity
toward acceptors that 7c-back-bond when a data set involves only phosphines. The ECW
approach and its interpretations of reactivity are contrasted with the reported
unprecedented conclusions from correlations using cone angles and one parameter basicity
Master Fit for Determining EB and CB for Phosphines
The measured physicochemical properties for all donors (Table 2-1) and all the
acceptors (Table 2-2) are substituted into Equation (2-7) leading to a series of
simultaneous equations. In most instances each equation has five unknowns. When the
EA and CA values are known from earlier studies,'1 these are entered into the equation and
held constant. When donors from the E and C correlation other than phosphines are used
in the study of a reaction or spectral shift, their EB and CB values'2 are also entered into
the equation and fixed in the data fit. The best set of unknown parameters are determined
by a least squares minimization'2 for the entire set of weighted simultaneous equations.
This is referred to as the master fit.
In using the master fit to determine phosphine a-donor parameters, it is essential to
eliminate contributions from both enthalpic and entropic steric effects from the data set.
Since the connection between the phosphine parameters and those of other donors is
critically dependent on systems like CF3SO3H, C6I5OH, and A12(CH3)6, which are
measured with different families of donors, a large weight was given to these acceptors in
the master fit. In general, enthalpies are assigned weight values of 1, 13C NMR shifts a
value of 1, redox potentials a value of 2 (in view of the small range of values that are
accurately known), and infrared shifts, Av a value of 0.1 (because the values span a large
range) in the data fit. The logs of rate constants are assigned weights of between 0.6 and
Table 2-1. EB and CB Parameters for Phosphines
no. phosphine wte EB CB CB/EB Ob
1 P(CH3)3 1.0 0.31 5.15 17 118
2 P(C2Hs)3 1.0 0.28 5.53 20 132
3 P(n-C3H7)3 0.5 0.37 5.16 14 132
4 P(i-C3H7)3 0.7 0.36 5.46 15 160
5 P(n-C4H9)3 1.0 0.32 5.36 17 132
6 P(i-C4H9)3 0.7 0.48 4.60 10 143
7 P(t-C4H9)3 0.5 0.25 6.08 24 182
8 P(c-C6Hn1)3 1.0 0.41 5.35 13 170
9 P(CH2CH2CN)3 0.5 0.95 1.50 1.6 132
10 P(CH2C6H5)3 0.2 0.63 3.32 5.3 165
11 P(OCH3)3 1.0 0.50 3.32 6.6 107
12 P(OC2Hs)3 1.0 0.56 3.17 5.7 109
13 P(O-i-C3H7)3 0.7 0.53 3.59 6.8 130
14 P(OC4H9)3 0.5 0.45 3.86 8.6 110
15 P(OC-Hs)3 1.0 0.71 1.69 2.4 128
16 P(OCH2)3R 0.7 0.09 4.85 54 101
17 P[N(CH3)2]3 0.3 0.38 5.11 13 152
18 P(CH2CH=CH2)3 0.2 0.64 3.61 5.6 c
19 P(C6H5)3 1.0 0.70 3.05 4.4 145
20 P(4-CH3C6H4)3 1.0 0.65 3.41 5.2 145
21 P(4-OCH3C6H4)3 1.0 0.62 3.57 5.8 145
22 P(4-FC6-4)3 1.0 0.74 2.70 3.6 145
23 P(4-CIC6H4)3 1.0 0.82 2.35 2.9 145
24 P(4-CF3C6-H4)3 1.0 0.91 1.52 1.7 145
25 P(4-NMe2C6H4)3 0.5 0.05 6.90 140 145
26 P(3-CH3C6H4)3 1.0 0.55 3.83 7.0 145
27 PCI3 0.2 0.70 0.18 0.26 124
28 P(CH3)2C6H5 1.0 0.44 4.49 10 122
29 P(C2Hs)2C6H5 1.0 0.39 4.91 13 136
30 P(OCH3)2C6H5 0.5 0.59 3.39 5.7 120
31 P(Cl)2C6H5 0.2 0.92 0.25 0.27 131
32 P(C6H5)2CH3 1.0 0.57 3.74 6.6 136
33 P(C6H5)2C2H5 1.0 0.55 3.83 4.4 140
34 P(C6H5)2 n-C49 0.5 0.58 3.80 6.6 140
35 P(C6H5S)20 OCH3 1.0 0.59 3.39 5.7 132
36 P(C6H5)2 Cl 0.5 0.66 2.35 3.6 138
37 P(C-6HI)2H 0.2 0.49 4.51 9.2 143
38 AsPh3 0.2 0.90 2.16 2.4 141
(a) If more than 12 systems are studied, a weight (wt) value of 1 is assigned, 12-10 a
value of 0.7, 9-7 a value of 0.5, less than 7 a value of 0.3. If all the acceptors
studied for a donor have CA/EA ratios that do not differ by more than 1.0 or if a
given phosphine has not been studied with at least one acceptor that also has
measurements with donors other than phosphines, 0.1 is subtracted.
(b) Cone angles are from refs. 23, 29, 32, and 49.
(c) No cone angle is reported in refs. 23, 29, 32, or 49.
Table 2-2. EA*, CA*, and W Parameters for Acceptor Systems
Acceptor Property wt. I EA* CA* w CA*/EA*
-AH (CF3SO3H) b 1.2 4.51 5.70 0.84 1.26
-AH (B(CH3)3)c 0.4 3.57 2.97 0 0.83
-AH (AI(CH3)3) d 1.2 8.28 3.23 -8.46 0.39
-AH (CpIr(CO)PR3)c 0.5 1.16 1.52 24.8 1.31
-AH (Cp*Ir(CO)PR3)f 0.1 36.68 7.11 -10.18 0.19
AH (Ti(C7Hn1)2PX3)5 0.4 -9.10 0.83 13.13 -0.09
-Al (HgCI2) h 0.1 9.94 6.75 -9.46 0.68
-AHI (HgBr2)h 0.1 21.13 8.01 -20.88 0.38
-AH ([Ni ri-CsH7)CH3]2)' 0.8 -72.6 -11.7 111.5 0.16
-AH (CpMo(CO)3CH3) J 0.1 -30.57 -1.75 38.49 0.06
-AH (CpMo(CO)3C2Hs) J0.1 -23.08 -0.97 34.55 0.04
AHt (CoNO(CO)3) k 0.1 1.44 -2.88 22.83 -2.0
AHI (Ru(CO)4PX3(Dis 1))' 0.1 3.97 0.77 26.4 0.19
AHl (V(CO)6 (SN2)) = 0.1 -8.16 -2.44 22.98 0.30
All (Ru(CO)3PX3(SiCI3)2)" 0.8 -0.68 0.47 26.02 -0.69
AHl (Rh2(OAc)4) 0 0.1 4.79 0.78 5.65 0.16
AHI (Ru6C(CO)1_) p0.1 -1.64 -3.21 19.15 1.96
AH: (Cp(CQ2Me)C(CO)2) q 0.1 -11.02 -0.79 22.00 0.07
13C (Ni(CO)3L) r 0.8 8.27 1.95 -7.47 0.24
13C (Cr(CO)5L)' 1.2 8.87 1.84 -6.67 0.21
13C (Mo(CO)sPX3)' 0.8 7.90 1.63 -5.82 0.21
13C (W(CO)sL) u 0.1 13.74 2.36 180.1 0.17
-3C (CpMn(CO)2L) V 0.1 -22.11 -3.01 -208.1 0.14
13C (PtPh2(CO)L)w 0.5 21.76 3.90 154.3 0.18
'H (CpMo(CO)2LMe)x 0.3 0.63 0.71 3.73 1.13
FH (CpMo(CO)2LCOMe) Y 0.3 1.66 0.28 3.04 0.17
'H ((CpCO2Me)Co)z 0.8 -1.04 -0.11 4.58 0.11
'H ((MePAr3)Br) 0.1 -4.60 -1.05 9.67 0.23
v (Ni(CO)3PX3) 'b 0.4 -52.4 -12.2 2143 0.23
v (Ru(CO)3L) 0.4 -72.2 -14.7 2164 0.20
v (CH3CpMn(CO)2PX3) dd 0.4 -15.0 -4.9 1967 0.33
v (rI-CpFe(CO)(COMe)PX3) 0.4 -114.9 -21.7 2069 0.19
v (ri-Cp'Fe(CO)(COMe)PX3) 0.4 -114.1 -21.6 2063 0.19
v (Rh(OAc)4L) ff 0.1 -7.64 -1.57 42.96 0.21
v [(Fe(CO)3(PR3)C7H9)+] U 0.4 -10.63 -3.60 2074 0.34
v (PtPh2(CO)L) hh 0.4 -86.6 -17.2 2177 0.20
v ((CpCO2Me)Co) 0.4 -60.8 -16.8 2034 0.28
v (Ru(CO)3PX3(SiCl3)2)'jj 0.4 -72.2 -14.7 2164 0.20
Acceptor Property wt. [EA* CA* W CA*/EA*
EI2 (Cp'Mn(CH2C12)) 1.7 -0.37 -0.12 1.16 0.32
E2 (Cp'Mn(CH3CN))" 1.7 -1.10 -0.25 2.05 0.23
El/2 [(Ru(bpy)2PX3)2+(CH3CN)] Im 1.3 0.30 -0.03 1.20 -0.10
Ei/2 [(Ru(bpy)2PX3)2(4Ac-py)]mm 0.1 0.11 -0.06 1.43 -0.55
EIr2 [(Ru(bpy)2PX3)2+(Cl')] m 0.1 0.28 -0.03 1.24 -0.11
EI/2 (1i-CpFe(CO)(COMe)PX3) 1.3 -0.89 -0.19 1.53 0.21
E1/2 (r|-Cp'Fe(CO)(COMe)PX3) 1.3 -0.85 -0.19 1.47 0.22
log K (CF3Ce-4OH) 0.6 1.69 0.26 -1.49 0.15
log K (CpMo(CO)2LCOMe)" PP 0.1 28.26 3.84 -34.55 0.14
log Ki (Rh2(OAc)4L) qq 0.1 4.76 1.30 -2.15 0.28
log K (W(CO)s(aniline)) '0.1 0.30 0.43 -0.20 1.43
logK (W(CO)s(p-Br-aniline))" 0.1 -1.64 -0.01 2.64 0.01
log k (Co(NO)(CO)3 (SN2))" 0.5 4.98 1.51 -10.71 0.30
log k (V(CO)6 (SN2)) 0.3 0.60 1.37 -5.10 2.28
log k (Ru(CO)4L) 0.8 2.77 0.41 -6.42 0.15
log k (CpzFe2(CO)4) w0.3 -1.28 -0.66 8.38 0.52
log k (CpMn(py)) 0.1 -13.75 -0.01 10.81 0.001
log k (Ru(bpy)2PX3(I20)2)xx 0.5 12.24 2.40 -18.14 0.20
log k (Co-C6HsCH2Br) Yy3 0.9 -4.38 0.43 -0.05 -0.13
log k (Co'-C6HsCH2Br) yy 0.4 -3.58 0.13 0.46 -0.04
log k (Fe(CO)2Cp(ethene)) z 0.1 24.34 5.01 -30.49 0.21
log k (ri-MesCpRh(CO)2) 0.1 -7.38 -0.23 0.90 0.03
log k (Mn(CO)2(NO)(6-C6H6)) bbb 0.1 0.26 0.99 -2.75 3.81
log k (CpIr(CO)PX3) 0.3 5.32 1.58 -10.08 0.30
log k (Cp*Ir(CO)PX3) 0.1 -1.94 0.46 -0.18 -0.24
log k (C2H5I) ddd 0.5 -0.63 0.46 -5.05 -0.73
log k (ni-MesCpCo(CO)2)' 0.3 -5.22 0.15 -1.88 -0.03
log k (Ti-CpMo(CO)2LCOMe) fr 0.3 10.60 1.56 -17.11 0.15
log k (Fe(I-Cp)(CO)LMe') U 0.8 -3.67 -0.72 3.61 0.20
log k (Fe(ri-Cp)(CO)LMeAN) 0.8 18.71 3.29 -23.03 0.18
log k2 (Ru6C(CO)17)" '0.3 -13.93 -0.87 11.55 0.06
log k2 (Mo2(CO)S)W 0.1 -2.53 -0.57 -0.41 0.23
log k (Fe(CO)3(C7H9)) "k 0.1 16.30 3.53 -19.96 0.22
log k (PtPh2CO(5-AQ)) n 0.8 -22.81 -3.33 26.24 0.15
log k (Os3H2(CO)0lo) '0.5 -2.92 0.59 2.01 -0.20
log T, [Re(CO)3Cl(PPh3)2]ax 0.1 1.98 0.39 -1.78 0.20
log Ti [Re(CO)3Cl(PPh3)2]equil 0.1 0.49 0.27 -1.22 0.55
pKA ow 0.8 22.03 5.81 -29.87 0.26
(a) The wt (weight) value is to be used in future correlations with this physicochemical
property to determine new donor EB and CB. If more than 12 donors give
satisfactory data fits, a wt of 1 is assigned, 10-12 a value of 0.7, 7-9 a value of 0.5,
and less than 7 a value of 0.2. If donors other than phosphines are fit, 0.2 is
added. If not, 0.2 is subtracted. In view of the small magnitude, 0.5 is added to
E1/2 values. A value of 0.4 is assigned to IR shifts because of their large
magnitude. Smaller wts. should be assigned to free energies (other than E1/2) for
substituents where entropic factors can contribute. The weight in a fit is related to
n used in articles before 1996 by wt=l/5n.
(b) -AH for the reaction of CF3SO3H with bases in 1,2 dichloroethane solvent.
x=0.07, % fit=0.3. Data from ref. 50.
(c) Gas phase -AH of adduct formation. x=0.32, % fit=7. Data from ref 51.
(d) -AH for the reaction of [Al(CH3)3] with donors in hexane solvent. x=0.26, %
fit=2. Data from ref. 52 with Et20 and EtaN omitted.
(e) -AH of protonation of CpIr(CO)PX3 with CF3SO3H in 1,2 dichloroethane.
x=0.26, % fit=5. Data from ref. 50b, 53 with P(c-hex)3 omitted.
(f) -AH of protonation of Cp*Ir(CO)PX3 with CF3SO3H in 1,2 dichloroethane.
x=0.22, % fit=5. Data from ref. 53b.
(g) Endothermic AH for PX3 dissociation from bis(2,4dimethylpentadienyl)titanium in
THF solvent. x=0.04, % fit=0.7. Data from ref. 54 with PEt3 omitted.
(h) -AH of 1:1 adduct formation in benzene. For HgCl2, x=l.10, % fit=8. For HgBr2,
x=0.61, % fit=5. Data from ref. 55 with P(c-hex)3, benzene, pyridine, and THF
(i) Heat evolved corrected for the heat of solution of the base in kcal mol' when a 1.0
M solution of the donor is added to 0.05 M di-t-methylbis[1-methyl- l-_3-(2-
butenyl)] dinickel in tetralin. x=0.06, % fit=0.4. Data from ref. 56 with P(OPh)3
and PPh3 omitted.
(j) Enthalpy for the insertion of CO in the Mo-R bond of CpMo(CO)3R, where R=Me
or Et and coordination of PR3 to form CpMo(CO)2(PR3)RC(O). For R=Me,
x=0.08, % fit=2. For R=Et, x=0.21, % fit=3. Data from ref 57.
(k) Activation enthalpy for the second order substitution of CO by phosphines for
Co(NO)(CO)3 in toluene. x=0.04, % fit=l. Data from ref 58.
(1) Activation enthalpy for the first order dissociative substitution of CO in
Ru(CO)4PX3 by P(OEt)3 in hexane and decalin. x=0.4, r2=0.92. A steric onset of
128 and an s of-0.12 is needed. Data from ref 59.
(m) Activation enthalpy for second order substitution of CO by phosphines in V(CO)6
in hexane. x=0.22, % fit=7. Data from ref. 60.
(n) Activation enthalpy for the first order dissociative substitution of CO in
Ru(CO)3PX3(SiCl3)2 by P(OMe)3 or P(t-C4H9)3. x=0.23, % fit=7. Data from ref.
(o) Activation enthalpy for the substitution of solvent with phosphines in dirhodium
(II) tetraacetate in CH3CN. x=0.20, % fit=17. Data from ref 62 with P(OPh)3
and P(benzyl)3 omitted.
(p) Activation enthalpy for the second order reaction of Ru6C(CO)17 with nucleophiles
in heptane. x=0.30, % fit=5. Data from ref. 63.
(q) Activation enthalpy for the second order substitution of CO with ligands in (1'5-
CsH4CO2Me)Co(CO)2. x=0.13, % fit=--9. Data from ref. 64 with all donors
omitted that have two or more phenyl substituents and P(OPh)3.
(r) 3C chemical shift of Ni(CO)3PX3 relative to Ni(CO)4 in CDCI3. Data from ref. 65.
(s) 13C chemical shift of the cis-carbonyl in Cr(CO)sL relative to Cr(CO)6 in CDC13.
x=O0.12, % fit=2. Data from ref 65 with pyridine omitted.
(t) '3C chemical shift of the cis-carbonyl in Mo(CO)sPX3 relative to Mo(CO)6 in
CDCl3. x=0.14, % fit=2. Data from ref 65.
(u) 13C chemical shift of the cis-carbonyl in W(CO)sPX3 downfield from TMS in
CDCl3. x=0.22, % fit=6. Data from ref. 66 with 4-methylpyridine omitted.
(v) 13C chemical shift for the CO in CpMn(CO)2PR3 relative to TMS in CDCI3.
x=0.26, % fit=6. Data from ref. 67.
(w) 13C chemical shift for cis-[PtPh2(CO)2L] relative to TMS. x=0.17, % fit=25. Data
from ref. 68 with P(i-Pr)3, P(t-Bu)3, and P(c-hex)3 omitted.
(x) 'H chemical shift for the Cp group in CpMo(CO)2LMe relative to TMS. x=0.04,
% fit= 20. Data from ref. 69.
(y) 'H chemical shift for the Cp group in CpMo(CO)2LCOMe relative to TMS.
x=0.04, % fit=16. Data from ref 69.
(z) 'H chemical shift for the CO2Me group in (il5-CpCO2Me)Co(CO)L relative to
TMS in CDCl3. x=0.05, % fit= 22. Data from ref 64 with P(c-hex)3 omitted.
(aa) 'H chemical shift for the methyl group in [MePAr3]Br relative to TMS. x=0.02, %
fit=2. Data from ref. 70.
(bb) A,, CO stretching frequency (cm') of Ni(CO)3PX3 in CH2C12. x=1.26, % fit=3.
Data from ref 71 with P(OCH2)3CR and P(O-i-Pr)3 omitted.
(cc) vi(ax), CO stretching frequency of Ru(CO)4L in heptane or hexane. x=l. 18, %
fit=7. Data from ref 59 with AsPh3 omitted.
(dd) Higher energy CO stretching frequency (cm-1) of IS-MeCpMn(CO)2PX3 in
heptane. x=1.31, % fit=19. Data from ref. 72 with nitrogen donors and PPhEt2
(ee) CO stretching frequency of i5-CpFe(CO)COCH3PX3 in cyclohexane. for Cp,
x=0.89, % fit=2. for Cp*, x=0.87, % fit=4. Data from ref 73.
(ft) Electronic absorption maximum in 10"3 cm'" for Rh2(OAc)4(CH3CN)L. x=0.39, %
fit=18. Data from ref 62 with AsPh3.
(gg) Higher energy CO stretching frequency (cm'1) for [Fe(CO)3(1-4-Tj-R3P.C7H9)]r
adducts. x=O.35, % fit=9. Data from ref 74.
(hh) CO stretching frequency for cis-[PtPh2(CO)L]. x=0.82, % fit=3. Data from ref.
68 with P(i-Pr)3, P(t-Bu)3, and P(c-hex)3 omitted.
(ii) CO stretching frequency of(ri5-CpCO2Me)Co(CO)L. x=0.93, % fit=3. Data from
ref. 64 with P(c-hex)3 omitted.
(ij) CO stretching frequency of Ru(CO)3L(SiCl3)2. x=0.82, % fit=3. Data from ref. 61
with P(OPh)3 and P(benzyl)3 omitted.
(kk) Standard oxidation potential in V of CH3CpMn(CO)2L in CH2C12. x=0.02, %
fit=5. Data from ref 72a.
(11) Same as (kk), but in CH3CN. x=0.02, % fit=3. Data from ref. 72b.
(mm) Redox potential for [Ru(H20)(bpy)2PX3]2+/'3 in CH2C2 vs. SCE. with an incoming
group ofCH3CN, 4-Ac-pyridine or C'. for CH3CN, x=0.02, % fit=4. for 4-Ac
-py, x=0.02, % fit=6. for CI, x=0.01, % fit=3. Data from ref 75.
(nn) Redox potential for i5-Cp and TS5-Cp*Fe(CO)(COCH3)PX3 in CH3CN (0.2 M
LiCO104) vs. SCE. for Cp, x=0.01, % fit=3. for Cp', x=0.01, % fit=3. Data from
(oo) Log of the equilibrium constant for PX3 hydrogen bonding to 4-CF3-C6H4OH in
CS2 at 25C. x=0.07, % fit=3. Data from ref. 76 with P(Ph-p-OCH3)3 omitted.
(pp) Log of the equilibrium constant for ligand dissociation from CpMo(CO)2LCOCH3.
x=0.14, % fit=5. Data from ref 69 with PBu3 omitted.
(qq) Log of the equilibrium constant for ligand addition to Rh2(OAc)4(CH3CN)2.
x-=0.73, % fit=23. Data from ref 62 with AsPh3 omitted.
(rrf) Log of the equilibrium constant for substitution of either aniline ofp-Br-aniline
with a ligand on W(CO)sL. for aniline, x=0.09, % fit=6. for p-Br-aniline, x=-0.13,
% fit=13. Data from ref 77 with AsPh3 omitted.
(ss) Log rate constant for the second order displacement of CO by phosphines from
Co(NO)(CO)3 in toluene. x=0.26, % fit=6. Data from ref 58 with nitrogen
donors, P(c-hex)3, and P(NMe2)3.
(tt) Log rate constant for the second order displacement of CO from V(CO)6 by PX3
at 25C in hexane. x=0.22, % fit=6. Data from ref 60 with P(i-Pr)3, PBu3, and
(uu) Log rate constant for the first dissociative substitution of CO from Ru(CO)4L by
another L to form Ru(CO)3L2 in heptane at 60C. x=-0.19, % fit=12. Data from
ref 59 with PEt3, PBu3, and P(c-hex)3 omitted.
(vv) Log rate constant for the second order addition of PX3 to Cp2Fe2(CO)3 in hexane
at 25C. x-0.05, % fit=3. Data from ref 79 with P(OPh)3, PBu3, and CH3CN
(ww) Log rate constant for the second order substitution of 4-NO2CsH5N in the electro-
chemically generated cation MeCpMn(CO)2-4-NO2C5HsN+ by phosphine and
other donor ligands in CH2C12. x=0.06, % fit=3. Data from ref. 72a with PEt3,
PPhEt2, PPh2Et, and PPh2Bu omitted.
(xx) Log second order rate constant for exchange of H20 by CH3CN in
RuU(bpy)2(PX3)(H20)2+ in o-dichlorobenzene. x=0.18, % fit=12. Data from ref.
75 with PEt3, PPr3, and PBu3 omitted.
(yy) Log rate constants for the reactions of Co(bis(dioximato)cobalt(ll)-L) and
Co'(bis-(1,2-cyclohexanedionedioximato)cobalt(II)L) with C-H5CH2Br in benzene.
for Co, x=0.20, % fit=8. for Co', x=0.13, % fit=7. Data from ref 80 with PEt3
and PBu3 omitted.
(zz) Log of the second order rate constant for phosphorus nucleophiles toward
[CpFe(CO)2(TI-C2H4)]+. x=0.26, % fit=9. Data from ref. 81 with P(CH2CH2CN)3
(aaa) Log of the second order rate constant for CO substitution by a ligand in Rh(Til5-
CsMes)(CO)2 in toluene. x=0.18, % fit=7. Data from ref. 82 with PBu3 and P(i
(bbb) Log of the second order rate constant for nucleophilic addition in Mn(CO)2(NO)(6
-MeCHI6) in CH3CN. x=0.09, % fit=4. Data from ref. 83 with nitrogen donors
(ccc) Log of the second order rate constant of CpIr complexes with CH3I in CD2C12 at
25C. For Cp, x=0.07, % fit=3. for Cp*, x=0.17, % fit=6. Data from ref 7b with
PEt3 and P(c-hex)3 omitted.
(ddd) Log of the second order rate constant between phosphine and C2HsI in acetone.
x=0.15, % fit=9. Data from ref. 84 with pyridine omitted.
(eee) same as (bbb), except with Co(i5-CsMes)(CO)2. x=0.11, % fit=7. Data from ref.
82 with PBu3 and P(i-Bu)3 omitted.
(ffi) Log rate constant for CO dissociation from CpMo(CO)2L(COMe) in CH3CN.
x=O0.10, % fit=9. Data from ref 69.
(ggg) Log of the second order rate constant for CO insertion for (TI-Cp)Fe(CO)LMe' in
CH2C12. x=0.10, % fit=l 1. Data from ref. 85 with PEt3 omitted.
(hhh) Log of the second order rate constant for the substitution of CO for CH3CN in
CpFe(COMe)L(CH3CN). x=0.36, % fit=20. Data from ref. 86 with PEt3, P(i
-Pr)3, P(i-Bu)3, and P(c-hex)3 omitted.
(iii) Log of the second order rate constant for reaction of Ru6C(CO)17 with
nucleophiles in chlorobenzene. x=0.35, % fit=7. Data from ref 63 with P(O-i-
Pr)3 and PBu3 omitted.
(jjj) Log of the first order rate constant of the dissociation of a ligand in [(CO)4Mo(A-
PEt2)2Mo(CO)3L] in decalin. x=0.13, % fit=21. Data from ref 87 with PBu3
(kkk) Log of the second order rate constant for the addition of phosphorus nucleophiles
to Fe(CO)3(C7H9)+ in acetone at 20C. x=0.35, % fit=9. Data from ref. 74.
(111) Log of the second order rate constant for the substitution of 5-aminoquinoline
with ligand in [PtPh2CO(5-AQ)] in toluene at 25C. x=0.36, % fit=12. Data from
ref 68 with P(i-Pr)3 and P(t-Bu)3 omitted.
(mmm) Log of the second order rate constant for the addition of nucleophiles to (g12-
H)20s3(CO)I0 in heptane at 30C. x=0.27, % fit=8. Data from ref. 88 with
P(OPh)3, P(NMe2)3, P(t-Bu)3, P(c-hex)3, P(benzyl)3, and AsPh3 omitted.
(nnn) Log of the axial and equatorial 31p relaxation times in CD2C12. For axial, x=0.02,
% fit=5. for equatorial, x=0.02, % fit=4. Data from ref 89.
(ooo) pKA values. x=0.66, % fit=6. Data from ref. 90 with PMe3 omitted.
0.2 depending on the estimated severity of steric contributions. Acceptors with bulky
donors, for which preliminary fits indicate that steric repulsions do exist, are omitted.
With acceptors that t-backbond, donors other than phosphorus donors are not included
The EB and CB parameters for phosphorus donors are reported in Table 2-1. Also
given in Table 2-1 are the weights that reflect their relative certainty based on the number
and type of different acceptor used to determine the parameters. Several of the
phosphines were studied with a very limited number of physicochemical measurements.
The reported parameters for these systems are given low weights in Table 2-1, and the
parameters should be redetermined as more data become available.
Results and Discussion
ECW Analysis of New Acceptors
The donor parameters from the master fit described above can be used to analyze
measurements of reactivity and spectroscopy for a new acceptor. A good definition of
parameters in any fit requires that donors and acceptors be selected in which there is a
large variation in the relative importance of the components of the reactivity scale. In
ECW fits, the CB/EB ratio of the donors employed should vary. Each measurement with
the new acceptor produces an equation in the series with the form of Equation (2-7). The
weighted series can then be solved by least squares minimization routines for the
parameters EA*, CA*, and W. A good fit for a wide range of donor types indicates that
the acceptor property is dominated by the same electronic factors that influence c-bond
strength. If donors other than phosphorus donors are included and a poor fit results, these
other donors are omitted. A good fit of the remaining phosphines and phosphites suggests
a 7r-backbonding acceptor.
A poor fit of only phosphorus donors indicates added complexity in the reaction.
Donors that deviate by 2.5 times the average deviation are omitted and the fit is
redetermined. A good fit usually results and one looks for patterns (steric effects,
incomplete complexation, etc.) in the donors that were omitted to reveal the complicating
factor. The fit is again redetermined omitting those donors expected to have contributions
from the suspected effect. A good fit at this point leads to a tentative assignment of the
complication. Omitted donors that are fit well (less than 2.5 times the deviation) in this
last fit are added back into the final fit.
Systems were selected to show how steric effects can be parametrized in the ECW
model. This is done with an enthalpy of reaction or activation by first running a
correlation of all available data using Equation (2-7). If steric effects are operative in the
system, a poor correlation will result. Donors with the largest cone angles are then
removed one at a time, and the correlation redone until a good fit of the remaining donors
results. Using this fit, the deviations are calculated for the donors that have been
removed. These deviations are plotted versus their cone angles and a least squares
regression is done. The slope of the line, s, will give the severity of the steric effect, while
the intercept divided by s will give OON, the cone angle above which steric repulsion
becomes operative in the system.
ECW vs. Substituent Constant Correlations
The EB and CB parameters from this data fit can be compared to those estimated
by summing the substituent constants by substituting these quantities into Equations (2-5)
and (2-6) and solving for sBE and SBc. A poor fit results with r2 values of 0.64 and 0.60 for
Equations (2-5) and (2-6) respectively. The prediction of EB and CB by the summation of
substituent constants does not produce as accurate a measure of basicity as solving for EB
and CB with a data set that contains phosphines and other donors. Problems could arise
with substituents saturating the inductive properties, i.e., three electron withdrawing
alkoxy substituents do not cause incremental changes for each substituent added. It is also
possible that the influence of the substituent is conformation dependent and bulky
substituents are locked into certain conformations when reaction occurs making the
substituent constant estimate of CB unreliable.
Why do good fits result30 when the AEX and ACx substituents were used to analyze
the phosphine systems of Table 2-2? Good fits result because only phosphine donors are
used in substituent constant correlations. The small range of CB/EB values enables the fit
to compensate for the small but significant deviations in additivity by adjusting the fdE and
dc values of the acceptors. This makes interpretation of the dE and dc values difficult, but
does not impact on the use of substituent constants to spot irregularities (e.g., entropic
and enthalpic steric effects) in the chemistry of a series of phosphine donors.
The similarity in the CB/EB ratios is a serious problem for substituent constant
correlations in general. Unless phenyl, alkyl and alkoxy substituents are studied to afford
the maximum variation in the CB/EB ratios of the family, an apparently good correlation
can be meaningless.
The above conclusions regarding the need to vary the CB/EB ratio in a data fit also
apply to fits to Equation (2-7) or to any basicity scale. When only phosphines with a
similar ratio are used to characterize a reaction or spectral change of an acceptor, a very
shallow minimum exists in the data set leading to a wide range in the magnitude of EA*
and CA* values that provide good data fits. Considerable error could result in the EA* and
CA* values that the least squares routine selects as the best fit parameters. At best, these
parameters can only be used to predict properties for other phosphines with similar CB/EB
ratios. At worst, if bonding contributions from effects other than o bond formation are
accommodated in determining the minimum, the resulting EA* and CA* parameters are
Determining the Existence of n-Backbonding
Acceptors in which 7r-backbonding is expected (e.g., spectral shifts of
Ni(CO)3PX3) gave good correlations using the same phosphorus donor parameters as
those for acceptors in which only o-bond interactions are involved (e.g., enthalpies of
reaction of donors with CF3SO3H and A2(CH3)6). The data fits for acceptors that can C-
back-bond are restricted to phosphorus donors while those with acceptors that only o-
bond may include phosphines, phosphites, sulfides, amines, pyridines, ethers, etc.
Equations can be derived to show that such a result is possible if the n-acceptor properties
of the phosphines and phosphites, EA" and CA", decrease proportionately with increasing
o-donor strength, EB and CB. Equations (2-8) and (2-9), where k and k' are
proportionality constants, give such a relationship.
EA" = EA" kEB (2-8)
CA' = CA" k'CB (2-9)
The EA" and CA" terms refer to the 7c-acceptor contributions of a phosphorus
donor with no o-donor strength. When Equation (2-1) or (2-7) is fit to a data set that
contains a n-backbond contribution, the resulting parameters, EA*m, CA*F, and Wm, for
the 7r-backbonding acceptors are given by Equations (2-10), (2-11), and (2-12). The EB"m
and CBaM terms represent the electrostatic and covalent backbond-forming tendencies of
the metal. The correlation parameters, EA*F1T, CA*F, and W'r are difficult to interpret
because the separate contributions from 7c- and o-effects in Equations (2-10) (2-11)
cannot be separated.
EA*l = EA* kEa (2-10)
CA*' = CA* k'CBxM (2-11)
WFrr = W + EAEBM + CA'CB"M (2-12)
In these instances, the value of the correlation of the physicochemical property is to detect
donor contributions other than a bond strength.
A good data fit to Equation (2-7) will result for acceptors that n-backbond as long
as the same proportionality constants in Equations (2-8) and (2-9) apply to all the donors
studied. This is the case for all the phosphines and phosphites studied. However, different
constants are needed for other families of donors, e.g., pyridines. Thus, when data for
other families are combined with phosphorus donors, good fits to Equation (2-7) will
result for acceptors that only o-bond, but not for acceptors that also ni-back-bond.
The ECW Interpretation of Phosphine Enthalpies
Several enthalpy changes of reactions and enthalpies of activation have been
reported for phosphorus donors. Because a linear free energy assumption is not required
for enthalpy data, enthalpies provide the critical test of ECW or linear free energy models.
The enthalpies of formation for A12(CH3)6 adducts52 are important data for they include
seventeen phosphorus, sulfur, oxygen, and nitrogen donors. The observed enthalpies
corrected for the enthalpy of solution of the base are used in this correlation. The
resulting W value is endothermic (-AH = -8.5 kcal mol"') and corresponds to one-half the
enthalpy of dissociation of the dimer in solution. Steric effects are evident in the Et20 and
Et3N adducts. The correlation is illustrated graphically in Figure 2-la.
The enthalpies of dissociation of phosphorus donors from bis (2,4-
dimethylpentadienyl) titanium,54 Cp'2Ti, are fit very well to EA* = -9.10, CA* = 0.83, W
13.13 (r2= 0.99, F = 4440) of Equation (2-7). X is positive AH and as a result, a
positive parameter is endothermic. W has the wrong sign for association of THF after the
phosphorus ligand is displaced. The W sign arises from changes upon protonation of the
metal-phosphorus ligand EA"'EBm and CA CB"" terms of Equation (2-12) and indicates
a n-back-bond contribution. Triethylphosphine, the largest cone angle phosphine studied,
deviates and is omitted from the correlation because of an enthalpic steric effect.
Enthalpies for a wide range of donors [phosphines, pyridines, and (C2Hs)3N]
reacting with CF3SO3H were measured53 using 1,2-dichloroethane as the solvent. This is
the first system treated by ECW where the products are ionic; leading to an intimate ion-
pair in this solvent. All donors fit very well (r2=0.99, F=25000) with an average deviation
15 20 U
5 -----I --I I I ---
5 10 15 20 25 30 35
Calculated and experimental enthalpies of reaction for (a) [Al(CH3)3]2 (0),
with Et2O and EtsN deviating, and (b) the protonation of CpIr(CO)PR3
(A), with P(c-hex)3 deviating.
of 0.07 kcal mol"'. The small W value (1 kcal mol') could result from a constant, minor
difference in the solvation of the ion-paired product and the reactants.
Enthalpies of protonation of CpIr(CO)PX3 by CF3SO3H in 1,2 dichloroethane
were fit earlier for five reported51b1'53 phosphines. Subsequently, the number of
phosphines has increased53b to eleven. ECW analysis of the larger data set still gives an
excellent fit (r2=0.9) as illustrated in Figure 2-lb. P(c-hex)3 was omitted and gave a
predicted value that is 1.7 kcal mol"' too large. This deviation is attributed to larger
enthalpic steric repulsion of the phosphine ligands in the protonated complex than that in
the neutral complex or to a steric effect that weakens the ion-pairing energy to CF3SO3 in
the product. The large exothermic W value is not anticipated in view of the
CF3SO3H/phosphine fit. In this case the W corresponds to the EA"EB and CA-'CBm
terms of Equation (2-12). The enthalpies53b for the protonation of Cp*Ir(CO)PX3 give a
very good fit (r2=0.98, x=0.23), with very different parameters from those calculated
above for the Cp complex. Only five phosphines with a narrow range of Ca/Ea ratios
were used with the Cp* complex so these acceptor parameters are tentative.
Enthalpies of formation for 1:1 adducts of [TI-C5H7Ni(CH3)]2 have been
determined by adding excess base to a solution of the complex in tetralin.10 The
correlation is very good with EA = -74. ,CA = -11.8, W = 113.1, r2=0.99 and x=0.21 as
shown in Figure 2-2. Earlier analysis of this system required the removal of PPh2Et and
P(benzyl)3 whose improved parameters now fit. Both P(C6H5)3 and P(OPh)3 had
enthalpies smaller than predicted by ECW and were omitted from the fit. Bulky
phosphines like P(i-Pr)3 with a cone angle of 160 are well behaved. The experimental
enthalpies are based on the assumption that the limiting reagent is fully coordinated. The
20 25 30 35 40 45
Calculated and experimental enthalpies of reaction of [Ni(7I-CsH7)CH3]2
(V), with PPh3 and P(OPh)3 deviating.
two omitted donors are the weakest studied, they probably do not completely complex all
of the nickel, and their reported enthalpies would be too small. In the absence ofn7t-
backbonding, the W value for this system is expected to correspond to the endothermic
cleavage of the dimer, but W is large and exothermic. This result is attributed to an
extensive 7E-backbond stabilization contribution in this system, i.e., W includes the
EAbEBM and CA"CBM terms of Equation (2-12).
The average deviation of calculated and experimental enthalpies in all the enthalpy
correlations in Table 2-2 is 0.2 kcal mol'1 or less. When large % fits (see the footnotes to
Table 2-2) arise, the range of enthalpies measured is small. The few donors removed from
the correlations are invariably the bulkiest donors which cause enthalpic steric effects. In
phosphorus donors, steric effects usually involve P(t-Bu)3, and to a lesser extent P(i-Pr)3
and P(c-hex)3. The adduct formation enthalpies leave little doubt about the applicability
ofECW to phosphorus donors, and the need for a dual parameter scale to describe their
Enthalpies of activation are reported for associative and dissociative substitution
reactions of several metal complexes with phosphines. Those in the former category
provide excellent correlation statistics but the number of phosphines studied are limited.
The results of typical associative reaction correlations are shown in Figure 2-3 and include
Co(NO)(CO)3 (r2=0.99, F=824), V(CO)6 (r2=0.96, F=26.8), and Ru6C(CO)17 (r2--0.98,
The enthalpies of activation for displacing CO with phosphine in toluene solvent by
a mechanism first order in Co(15-CsH4CO2Me)(CO)2 and phosphine are reported" for
twelve phosphines. Large deviations occur for the six phosphines that contain two or
*t I I l I
6 8 10 12 14 16 18 20
Calculated and experimental activation enthalpy for (a) Ru6C(CO)17 (0)
and (b)CoNO(CO)3 (A).
more aromatic substituents. While these ligands have large cone angles, steric effects are
not indicated in this system because P(c-hex)3 is well behaved in the correlation (Figure 2-
4) when aromatic phosphines are omitted (r2=0.92, F=44.3). Aromatic solvents are
known12' to undergo n-complexation with aromatic donors, and loss of this interaction in
the transition state would account for the observed increase in activation energy above
that predicted. The reported frequencies, Vco, for the (ri5-CsH4CO2Me)Co(CO)L adducts
of all the phosphines studied give the excellent correlation shown in Figure 2-4 (r2=0.99,
F=280). This is a meaningful result because if 7r solvent interactions are the cause of the
activation enthalpy deviations, they are not expected to have a significant effect on Vco.
For an associative mechanism, the W value contains the energy needed to
dissociate a carbonyl when a phosphine with no basicity is involved in the transition state.
The transition state stabilization during nucleophilic attack should lead to a W value that is
larger (more endothermic) than the AH' value with any donor. This is found in all four
systems. Thus, the metal-phosphine interaction involves mainly o donation. The
activation enthalpies for the CO dissociation from Ru(CO)4PX3 for seven phosphorus
ligands are poorly fit (r=0.39) with an average deviation of 1.80 kcal mol1. The fit still
was not satisfactory with the bulky P(t-Bu)3 removed and this acceptor will be discussed
ECW Parametrization of Steric Effects
The poor correlation of the enthalpy of activation of Ru(CO)4PX3 to ECW
suggests that this data set could be used to test the addition of a steric term. The literature
and data fit ofB(CH3)3 enthalpies to ECW,12' indicate steric repulsion is absent with
I I II
Activation enthalpies (A) and (2034-vco)/10 (1) for Cp(CO2Me)Co(CO)2.
The W value is 2034 cnfm'1.
ammonia and primary amine donors, marginal with secondary amines, and appreciable
with tertiary amines. This pattern suggests that the enthalpic steric term should have the
form s(O OoN)8 where 0 is the donor cone angle, ON the cone angle for the onset of
repulsion, and s is the coefficient indicating the severity of the effect. The Kronecker
delta, 8, is zero when 0 < 9oN and one otherwise. This term differs from the linear steric
term of QALE and has the same form as their nonlinear component of the steric effect.91
The three smallest cone angle ligands for Ru(CO)4PX3 are fit to ECW. Adding any other
fourth ligand gives an unsatisfactory fit. The plot of the deviations of the excluded donors
from the three phosphorus donor fit versus 0 gives an intercept whose magnitude is sOoN
(dev = sO S~oN = sO intercept). A OoN value of 128 results from the slope and
intercept. This value is substituted into Equation (2-13), and the acceptor coefficients
obtained from a least squares data fit are EA* = 3.97, CA* = 0.77, s = -0.12, and W= 26.4
with r2= 0.92 and an average deviation of 0.46 kcal mol1 for all seven phosphines.
AX = EA*EB + CA*CB + W + s(O OON)8 (2-13)
The solid line of Figure 2-5 shows the fit of the calculated activation enthalpy from the E,
C, and steric term correlation to the experimental values, and the dashed line and triangles
show the fit of only the ECW contribution to the enthalpy. The ratio of unknown
parameters to data is large, but the conclusion that steric strain is relieved in the transition
state is valid. This system illustrates the use of a steric term that is compatible with the
integrity of the parameters in the ECW model. It should be emphasized that we are not
advocating a four parameter analysis of data. Instead, when ECW gives a poor correlation
for bulky acceptors, we are offering a method to determine quantitatively if an enthalpic
31 ..... -- ------ ---
27 .. P(t-Bu3 =
27 -" P(c-hex)3
25 1 1 1 ' i
25 26 27 28 29 30 31 32 33
Figure 2-5. ECW (A) and ECW + cone angle fits (0) for the first-order dissociative
activation enthalpy of CO from Ru(CO)4PX3.
steric explanation is reasonable. For most of the acceptors in this article too few of the
donors studied deviate to provide this check.
Poor correlation of the enthalpies of adduct formation in trans-
(CH3Pt[P(CH3)2C6H5]2L)+(PF6) complexes92 to ECW were obtained in this and the earlier
study. After removal of nitrogen donors and bulky phosphines ( 0 > 160 ), the r2 in the
master fit is 0.42, with an average deviation of 0.94 kcal mol"'. Large misses are seen with
bulky phosphines, but adding a steric term, Equation (2-13), does not produce a
satisfactory fit even with ON equal to zero. Large contributions from steric effects are
evident, but other complications exist in this data.
Interpretation of Spectral Shifts and Redox Potentials
The 13C chemical shifts of Ni(CO)3PX3 relative to Ni(CO)4 and the CO stretching
frequencies of these adducts are fit very well with EB and CB, (x = 0.09 and 1
respectively). Decreasing the formal charge on nickel by a more basic a bonding
phosphine increases Ni-CO 7r-backbonding which in turn decreases the C-O stretching
frequency. On the other hand, 7t-backbonding from nickel into the phosphine decreases
electron density on nickel and increases the C-O frequency. The zero valent nickel atom is
expected to be involved in 7t-backbonding to the phosphines. The excellent data fit
supports compensating a- and 7r-effects in the complexes, Equations (2-10) and (2-11).
As expected, W, the shift for a donor with EB=CB=0 is larger than any phosphine adduct
frequency. However, expected 7r-backbonding makes interpretation of EA*, CA* and W
difficult and limits their application to phosphines.
The qualitative interpretation of the trends in 5 "3C of Ni(CO)3PX3 is complicated
by changes in the 13C electron population from a and r effects as well as the influence of
those effects on ground and excited state energies.65 The more basic phosphine gives rise
to a larger 13C shift. The excellent data fit again indicates that compensating x and a
changes exist, i.e., Equations (2-10) and (2-11) apply. This compensation permits the use
813C of Ni(CO)3PX3 as a one parameter scale of phosphine o-basicity for physicochemical
properties with a CA*/EA* ratio of~0.1.
The same trends as in Ni(CO)3PX3 are also noted in the fit of the 13C chemical
shifts of Cr(CO)sPX3,W(CO)sPX3 and Mo(CO)sPX3 adducts. The '3C shift has been
reported for Cr(CO)spy and W(CO)54-CH3py. In both instances, the pyridine donor had
to be omitted from the fit because the 13C calculated from EA*, CA* and W is considerably
larger (~5 ppm) than measured. This deviation provides strong support for 7r-back-bond
contributions to the shifts and, as one would expect, indicates that the same k and k'
cannot be used in Equations (2-10) and (2-11) for both phosphines and pyridines.
The CO stretching frequencies for Ru(CO)4PX3, Ti'-Cp and il5-Cp'
Fe(CO)(COCH3)PX3 fit very well, and there is no indication of a steric contribution in any
of the compounds. Large cone angle phosphines were not studied with the latter two
systems, but were with the nickel and ruthenium complexes. None of the spectral shifts
show any indication of steric strain in the ground states of the complexes studied.
Fits of the reduction potentials for series of M-L complexes in which L is varied
measure the free energy of interaction of L with M in the oxidized and reduced forms of
the complex. The magnitudes of the parameters would be influenced by 7t-backbonding if
the interaction differs in the two oxidation states. Steric effects would cause deviations in
the data fit to ECW only to the extent that they differ in the two oxidation states. Average
deviations of 0.02 or better are obtained for all systems studied.
Interpretation of Reaction Rates
In contrast to the good data fits for most of the systems which involve enthalpies
of adduct formation, enthalpies of activation, spectral shifts, and E1/2, the log of the rate
constants are often poorly fit. Usually certain phosphines must be omitted from the
analysis. In the substituent constant analysis ofphosphine reactivity,30 deviations were
also found in the rate data that were not found in E1,2, spectral shift, and enthalpy analyses
of the same complexes. Two types ofsteric effects were suggested.30 The first is a cone
angle, front strain effect that is manifested in both the enthalpy and free energy of
interaction. The second was an entropic steric effect, often found in ethyl and longer alkyl
chain phosphines, that involved loss of rotational freedom in the chain in the course of
forming the transition state. These same patterns are found in the EB and CB fit of rate
data for most of the systems in Table 2-2, and the reader is referred to the earlier
literature30 for a discussion of the specific systems. In future analyses of free energy data,
a percent fit of greater than 6 (if experimental error warrants a better fit) would suggest
assigning less weight to the longer chain phosphine or adding an s(0 OoN)5 term to
determine the influence of entropic and enthalpic steric effects, respectively, on the
measurement. If these effects are not operative, the data fit will not be improved by the
omission of all long chain phosphines or addition of an s(O 9oN)5 term. Clearly, entropic
and enthalpic steric effects would require different parameters. Rather than trying to
parameterize entropic effects, it has been our philosophy to use ECW to detect
complications in these systems that are not related to donor-acceptor bond strength, and
to use other measurements to confirm the cause of deviations.
The analysis of free energies of reaction with parameters related to bond strength
requires a linear free energy assumption. This assumption is less likely to be valid on
complex inorganic and organometallic systems than on the structurally similar organic
systems treated with Hammett or Taft parameters. Six of the acceptors in Table 2-2 for
which both free energies and enthalpies are reported show unacceptable linear free energy
relations, i.e., entropy changes are not linear with enthalpy changes.
Comparison of ECW and Literature Analyses of Phosphine Reactivity
Typically, phosphorus donor reactivity has been analyzed with a one parameter
basicity scale, X, and a linear steric contribution, bO. These are linear free energy, bond
strength related parameters whose meaning is different than the parameters of the ECW
model. A X scale, based on the "3C or C-O frequency shift of Ni(CO)3L, is a one
parameter, basicity scale that applies to electrostatic acceptors with a CA/EA of 0.2 (see
Table 2-2). If the 0 term paralleled covalency, X/0 would fit phosphorus donor reactivity
as well as ECW. A recent "examination" of the ECW parameters with QALE claimed91
that EB and CB are linear combinations ofX and 0 with a small contribution from E,, "a
phenyl effect, whose origins are poorly understood." Such a result would have profound
implications for it would suggest, for phosphorus donors at least, that hard/soft,
covalent/electrostatic, and charge/frontier control are in effect hard/steric,
electrostatic/steric or charge/steric for any donor without a phenyl group. We would not
know if covalency, softness, or frontier control are in reality a linear steric effect.
The claim91 that EB and CB are linear combinations of mostly x and 0 parameters,
will be tested using enthalpies of reaction and activation. Enthalpies best measure bond
strengths and avoid a linear free energy assumption. 12' For enthalpic systems, ECW uses
only two parameters to fit the measurements, so the two QALE parameters x and 0 were
used in the comparison with the "minor Er" contribution9!' omitted. Enthalpies of reaction
for phosphorus donors with CF3SO3H, enthalpies of protonation of CpIr(CO)PX3,
enthalpies of adduct formation with Cp'2Ti and [Ti-CsH7Ni(CH3)]2, and enthalpies of
activation for associative CO substitution by phosphorus donors in CpCO2MeCo(CO)2
were fit to X and 0 to test their "equivalency" to the ECW fits described above. To afford
a direct comparison, those donors eliminated from the ECW fit because of steric effects
were also omitted in the QALE(x,0) fits. The enthalpies ofprotonation ofphosphines by
CF3SO3H are fit with an r2 of 0.98 using QALE(X,0). A poor x/0 fit results for [q1-
CsH7Ni(CH3)]2 (r2 is 0.81 with cz= 0.41, co=-0.068 and an intercept of 32.0). The Cp'2Ti
fit is not much better (r2 is 0.86 with cx=-0.52 co=-0.39, and an intercept of 65.7). The
protonation of Cplr(CO)PX3 fits very well (r2=0.98 with cx=-0.31, co=-0.056 and an
intercept of 42.4). The (O,x) fit of the enthalpies of activation for CO substitution by
phosphine in Co(ij5-CsH4CO2Me)(CO)2 give a poor fit (r2=0.72 with c,=-0.082, cO= -
0.035, and an intercept of 19.2). ECW correlated all these measurements to at least an
r2of 0.9. Clearly, the two parameter sets are not equivalent and the success of ECW
* In Equation 15 of a recent QALE article,91 an incorrect ECW fit of the enthalpies of
reaction of only phosphine donors with CF3SO3H is reported. The correct fit
parameters (using a previously reported set of EB and CB for phosphines) should have
been EA* =7.83, CA = 6.17 and W= -6.22 and the r2 is 1.0. The current phosphine
parameters give slightly different values.
indicates that the 0 term is added to x in an inadequate attempt to compensate for changes
in the covalency of the acceptor from that of the one parameter X scale.
Our earlier report detailing the inadequacy of one parameter basicity scales19"
predicts that the next step to be taken to improve data fits is to divide the data set into
subsets that limit the CB/EB ratios of the donors. As mentioned above, this leads to great
correlations with poorly defined parameters.21'47b When x/0 give poor correlations, this
subdivision of the data is accomplished in QALE by a variety of procedures: eliminate
phosphites, treat phenyl substituted phosphines separately, divide the set according to 0,
and treat only C3, phosphines.24'91'93'94'95 The deviating systems become "effects" that are
then parametrized empirically and used as needed in subsequent QALE analyses. It is not
surprising that one can derive parameters, e.g., the aryl effect,27 E., and steric
thresholds,95 to make these subsets conform and fit phosphine reactivity trends for many
ECW is criticized91 with the obvious claim that no two parameters will correlate all
phosphine reactivity. It is emphasized'2' that when Equation (2-7) is used to analyze data,
the conclusion is that the property either does or does not parallel bond strength. No
claim has ever been made to fit every property. The parametrization that leads to the extra
QALE parameters give better fits of free energies than ECW. However, what does it all
mean? The aryl effect parameter is not defined and its existence cannot be determined a
priori. To quote94 "we do not understand the nature of the aryl effect" and "the
contributions of E,, relative to x change from system to system and even change sign."
Adding a steric threshold suggests that there are two kinds of enthalpic steric effects.
What do presumably electronic parameters mean when they are able to fit non-linear free
energies? One is also prompted to ask, why are phosphorus donors so different from
other donors where these effects are not used? Another question seldom addressed in
QALE is "do the added parameters produce acceptor coefficients from the analyses that
make sense when different acceptors are compared?"
Meaning is clear in the ECW interpretation of the reactivity in these systems. The
parameters describe varying electrostatic and covalent contributions with a few deviations
attributed to steric effects. Explanations of this sort have precedence in the qualitative
explanations of reactivity. In QALE analyses, steric effects dominate phosphine reactivity,
with the coo steric term usually required in the analyses. The QALE fit of-AH for
CF3SO3H to phosphorus donors is excellent, giving fit parameters91
ofX = -1.05, 0 = -.088, and E,= -1.21 with an intercept of 51.3. Keeping in mind that 0
is usually ten times larger than x, a substantial steric effect, (coO), is involved for the
enthalpies of protonation by CF3SO3H according to these coefficients. The steric enthalpy
component for PMe3, for example, is 10 kcal mol'. It is hard to imagine a contemporary
view of reactivity to account for a repulsive enthalpy term of this magnitude toward the
proton that is not present in Ni(CO)3PMe3. Front strain steric effects in Bronsted acids
are without literature precedent and the QALE conclusion9r' that the coefficients from x/0
analyses are "consonant with the contemporary views" is not supported by this system.
The contribution of this steric component is not changed significantly when all the QALE
parameters are used.
With decreasing values of x for increasing basicity, and increasing values of 0 for
increasing steric repulsion, the ratio of the coefficients for fitting enthalpies of reactions
should be positive with the signs on each depending on the intercept. QALE parameters
often are not consistent with these signs, e.g., for [rI-CsH7Ni(CH3)]2 and apparently have
an unexplained meaning that is different than the contemporary view. These
unaddressed24"'9' questions and the inconsistencies described above must be answered to
accept the claim that the "QALE conclusions are consistent with contemporary views of
How does ECW explain the good fits that sometimes result if the QALE model is
using incorrect basicity parameters? For all phosphines and phosphites in Table 2-1 whose
weights are 0.7 or more, an excellent correlation ofX to EB and CB is found (r2 =0.98,
F=606 and x=1.0). This is expected because ECW gives an excellent fit to the frequency
shifts used to determine the X parameters. The C/E ratio for the x fit is 0.20 and that for v
Ni(CO)3L is 0.23. Any acceptor with a CA/EA ratio of 0.2 will correlate to ECW and to x
without need for a linear steric term even for different size acceptors. Next, the 0 term
was fit to EB and CB. When all of the donors used in the x fit were correlated, an r2 of
0.61 resulted. Fits were attempted with only phosphites removed and another with only
aromatic phosphines removed. The r2 values showed insignificant improvement. Thus,
there is no combination of Ea and CB that will be equivalent to 0 for these combinations of
donors. The omission of all but alkyl-substituted phosphines in the 0 fit led to the
excellent correlation ( r2 of 0.93) given in Equation (2-14).
9cALC = 399.7EB + 85.1 CB -446.9 (2-14)
The 0 values calculated using Equation (2-14) are plotted vs. literature values in Figure 2-
6 with the solid line drawn for the alkyl phosphines. When only alkyl phosphines are
studied, X and 0 will fit data as well as EB and CB but parameter meaning will differ. The
points in Figure 2-6 for other phosphines are not intended to give revised 0 values but to
show what the value of the cone angle would have to be to correlate to E and C. For
some donors, a conceptually impossible negative 0 would be needed. Since cone angles
are fixed, QALE adds new effects to bring the deviant points to the solid line and to do the
job EB and CB do. As shown in Figure 2-6, compounds containing a phenyl group show
negative deviations. With the PhPR2 deviating slightly, Ph2PR deviating to a greater
extent, and the average of the triphenyl substituted phosphines deviating the most, the plot
suggests that the "electronic aryl effect"'27 is just an added empirical parameter to
compensate 0 for its improper estimate of covalency in aromatic phosphines. Data can be
fit with this compensation but it is not surprising that meaning is lost.
Phosphites were not included in the recent comparison91 of QALE and ECW.
Figure 2-6 suggests that when a large number of donors are studied another effect, the
phosphite effect proposed in earlier QALE reports,73 will be needed to adjust for the
inadequacy of the cone angle to compensate for covalency in correlations of these donors
to acceptors that have a C/E ratio different than that of ECW recommends the use of
phosphites in all reactivity studies for they have a different CB/EB ratio than alkyl
Finally, Figure 2-6 provides the E/C explanation of why good QALE fits result
when a limited number of phosphorus donors are selected for study. If the donors
selected can have a line drawn through them in Figure 2-6, an equation similar to Equation
(2-14) can be derived to relate 0 to EB and CB. A good fit to QALE(x,0) will result for
this set of donors with the correlation coefficients adjusting to incorporate the new line
slope. The resulting coefficients will be without meaning. Two of many possible donor
1001 -- U l-
g 60 .-
I 40 oa .---
U 20 -
0 - -------- I------------I ----------'--I--
100 110 120 130 140 150 160 170 180 190
Cone Angle (literature)
Figure 2-6. Donor choices for which the cone angle, 0, is linear with EB and CB. The
solid line is a fit of the aliphatic-substituted phosphines to Equation (2-14).
Phosphines on the line will fit data to X and 0 as well as to EB and CB.
Points marked with 0 are aliphatic-substituted phosphines, those with X
are for phosphines with one aromatic group, those with U are for
phosphines with two aromatic groups, those with A are phosphines with
three aromatic groups, and those with are for phosphites. The dashed
lines represent arbitrary donor selections that would also be linear to EB
selections that can have a line drawn through them are illustrated by the dashed lines in
Figure 2-6. For these selections, good fits to x and 0 can result and if deviations occur for
a few of the selected phosphorus donors that are not on the line, the misses can be
accommodated by using the extra QALE terms. The signs and magnitudes of these terms
will not have meaning and will vary with donor selection to compensate for the 0
deficiency. The line drawn with large dashes will not require a phenyl effect but will need
a OON for large cone angles. Figure 2-6 should be used in phosphorus donor selection to
probe the steric contribution and can explain the need for extra QALE terms to make up
for covalency in data fits that are correlated by ECW with only two parameters.
ECW analyses of the reactivity and spectroscopy of substituted phosphines have
shown that spectral shifts, enthalpies of reaction and activation, and redox potentials are
primarily dominated by the donor strength of the phosphines and phosphites. Over 500 of
these physicochemical measurements are correlated to within experimental error with just
two sets of EB and CB parameters for phosphorus donors. Only in crowded acceptors
with bulky phosphines, are enthalpic steric effects evident. An extension of ECW is
offered using cone angles to verify the magnitude and onset of steric effects.
Rate constants for complex organometallic systems often do not exhibit linear free
energy behavior and thus should not be expected to correlate to enthalpic bond strength
parameters. Deviations are attributed to entropic steric contributions that result from
changes in substituent chain organization in the transition state or intermediate that are not
cone angle-related. These deviations are found in rate data but not in redox potential,
frequency shift or enthalpy data analyses of the same acceptor.
Analyses of phosphorus donor data with QALE procedures provides a very
different interpretation of reactivity than ECW. With QALE, steric effects are ubiquitous,
and no account is made for changes in covalency or softness of the acceptors. Hard/soft,
electrostatic/covalent or charge/frontier are replaced by hard/steric, electrostatic/steric or
charge/steric. In ECW, the importance of covalency is found to vary with the acceptor
and a linear steric contribution is not observed in most systems. Using the two QALE
Xand 0 parameters, it is shown by using data fits that these parameters are not "the same
as EB and CB except for a minor E., contribution."9' It is also shown that additional
empirical QALE parameters are used to obtain good fits of individual data sets at the
expense of meaning. This is evident in seldom made comparisons of the parameters for
QALE has been parametrized to correlate phosphine reactivity and will fit many
properties but in using the parameters one must be careful not to confuse correlation and
meaning. For example, what meaning can be inferred from the parameters when the
QALE linear free energy parameters fit non-linear free energy data sets? Is there meaning
in an unprecedented 10 kcal molP1 steric repulsion contribution toward the proton of
CF3SO3H when there is none for Ni(CO)3L? Other examples are given to show that the
meaning of the QALE coefficients do not represent contemporary views of reactivity. If
more attention is paid to the interpretation of coefficients from QALE fits and comparison
of these coefficients for different acceptors, more questions will arise about the procedure.
Finally, an ECW analysis of 0 indicates that the meaning of the extra terms of QALE are
to compensate for the failed attempt of 0 to accommodate the widely accepted differences
in covalency (softness) that exist in different acceptors.12"
In the larger scheme of understanding the factors that influence reactivity, the test
of any set of multiple scale reactivity parameters is not in slight differences in good data
fits. A more important concern is the consistency of the interpretation of the fit
coefficients when patterns for donor reactivity are compared for different acceptors. This
consistency affords understanding in the context of the model, leads to significant
generalizations of reactivity principles, and inspires new experimentation. Lack of
consistency, e.g., the E. parameter, suggests meaningless correlations with only limited
value for use in interpolative predictions.
EXTENSION OF THE ELECTROSTATIC-COVALENT MODEL TO 2:1 ADDUCTS
Work from this laboratory'2' has shown the utility of the Electrostatic-Covalent
Model (ECW), Equation (3-1), for the interpretation of a wide range of physicochemical
properties such as enthalpies of interaction, NMR, UV and IR shifts, E1/2 values, rate
constants, and activation enthalpies.
AX = EAEB + CACB+W (3-1)
AX is the magnitude of the physicochemical property (with units of energy), EA and CA
reflect the electrostatic and covalent properties of the acceptor, and EB and CB parallel the
electrostatic and covalent properties of the donor. W is the value of the physicochemical
property when EB=CB--O, and as an intercept it contains any contribution to a data set that
is independent of the base.
Recent additions (Chapter 2 and reference 30) to the ECW database have
significantly increased the relevance of the model to organometallic chemistry by
reporting EB and CB parameters for phosphines and phosphites. The electrostatic and
covalent phosphorus donor parameters are determined from data sets that are free of
steric effects. When deviations in ECW correlations indicate a steric effect, a procedure
is reported (Chapter 2) to confirm these effects quantitatively by adding a cone angle
term, s(O-9oN)5, to Equation 3-1. Here 0 is Tolman's cone angle, and 9oN is the cone
angle above which steric effects become operative. ECW analyses indicate that steric
effects play a greatly diminished role in understanding phosphine reactivity than suggested
by cone angle, 0, and one parameter basicity, x, models (Chapter 2 and reference 91).
The overestimate of steric contributions to reactivity with (X/0) is attributed to the failure
of any one parameter basicity scale to allow for variation in the covalency (softness) of the
different acceptors. In effect, 0 is used as an improper parameter for covalency and in
those instances where a successful (x/0) correlation does result, the acceptor fit
parameters are without meaning. In the many instances when (x/0) correlations fail, new
parameters are invented, such as the aryl effect (E.).27
Enthalpies for the reactions of two (or more) donors with a single acceptor or
physicochemical properties of adducts with two donors varied usually only allow
measurement of the total change for coordinating both donors. Equation 3-1 is not
expected to correlate total enthalpies for 2:1 donor-acceptor adduct formation. In the first
step of this type of reaction, each base forms a different 1:1 adduct which behaves as an
acceptor with different EA2 and CA2 parameters in the second step to form the 2:1
adduct videe infra).
A model has been derived96'97 for prediction of the enthalpies of reaction of two
bases with metal-metal bonded carboxylates of the general form M2(RCO2)4. This
equation is extended to the physicochemical properties for 2:1 adducts of acceptors in this
article. In order to correlate properties for 2:1 adducts with the ECW model, the two
steps are considered separately, as shown in Equations (3-2) and (3-3), where A is the
acceptor and B is the same donor for both steps.
A + B AB (3-2)
AB + B -> AB2 (3-3)
The enthalpy for step one [Equation (3-2)] is given in Equation (3-4), for reactions where
-AHI = EAEB + CACB (3-4)
After one donor molecule has been added, the EA and CA parameters will change for the
coordination of the second donor molecule. Strong donors are expected to produce a
weaker acceptor for the second step than weak donors because of the partial positive
charge remaining on the acceptor center. Relationships that express this behavior are
given in Equations (3-5) and (3-6), where EA2d and CA2d are the acceptor parameters for
the 1:1 adduct that reacts in the second step and k and k' are proportionality constants
characteristic of the acceptor that indicate the extent to which base coordination, kEa and
k'CB, modifies the acidity for the second step.
EA2 =EA- kEB (3-5)
CAd = CA k'CB (3-6)
Therefore, the enthalpy of reaction for the addition of the second donor to AB is given in
Equation (3-7) where EA2d and CA2d have different values for each base.
-AH2ld = EA2dEB + CA2ndCB = (EA kEB)EB + (CA k'CB)CB (3-7)
Combining Equations (3-4) and (3-7) gives the total enthalpy for the two steps in
Equation (3-8). Using the reported (Chapter 2) EB and CB parameters for the donor
molecules, a least squares minimization routine can be used to calculate the best acceptor
EA, CA, k, and k' parameters for the total enthalpy.
AHIT = -(AHI + AH2) = 2 EAEB kEB2 + 2 CACB k'CB2 (3-8)
Because four unknowns need to be determined, a large data set of donors with widely
varying CB/EB ratios must be studied to obtain a satisfactory solution for the unknowns.
When the 2:1 discussion for o acceptors is extended to acceptors that can undergo
n-back bonding, the Co-r relationships shown in Equations (2-10) and (2-11) are expected
to apply for phosphorus donors. Correlations can result for phosphorus donors, but the
parameters will contain 7c and o-effects that cannot be resolved at present.
The extension of the enthalpy discussion given above to physicochemical
properties that arise from coordination of two donors lead to Equation (3-9).
AXT = AXI + AX2 = 2 EAEB kEB2 + 2 CACB k'CB2 + W (3-9)
In this article, successful correlations of 2:1 adduct properties are reported, and the
interpretation of the correlations are discussed.
For a given acceptor, the acid in the second step will have an acidity that varies
depending on the base attached. As a result there is no average acid parameter that can fit
the total enthalpy. All reported literature analyses of 2:1 adducts, tacitly assume that the
two steps can be described with an average acceptor parameter when the total
physicochemical property is fit to a single acceptor parameter. Before any set of donor
parameters can be meaningfully applied to the analysis of 2:1 adducts, it is essential to
illustrate mathematically how the equation employed takes into account the changes
described above for the 2:1 adducts. The random fitting of 2:1 data to a set of parameters
derived for 1:1 adducts may yield a good fit but will lead to meaningless acceptor
parameters unless a derivation is provided to accommodate two different acceptors in each
Basicities ofbidentate donors have not been treated with ECW. Coordination of
the first donor atom is treated with Equation (3-1) using constant EB and CB values
characteristic of the donor reacting as a monodentate. As with 2:1 acceptors,
coordination to the first donor atom of the bidentate to an acceptor modifies the donor
properties of the second donor atom to a different extent for different acceptors. There is
no constant EB and CB for the second step if acceptor coordination has an inductive effect
that influences basicity. In this article, equations are derived for the analysis of these
The nitrogen donor EB and CB parameters used in these analyses are listed in Table
3-1. EB and Ca values for phosphorus donors are listed in Table 2-1. Physicochemical
measurements, listed in Table 3-2, were analyzed using the NCSS 5.X (Kaysville, UT)
statistical software. Regression analyses were done with the multiple regression module
of the software. The measured values for the donors were weighted as in Chapter 2 for
phosphorus donors and reference 12a for all other donors. If one uses EB and CB as two
of the independent variables [as in Equations (3-8) and (3-9)], their coefficients from the
regression are 2EA and 2CA, respectively, and must be divided by 2 to give the EA and CA
values. The other two independent variables that are used, E 2 and Ca2, will give
coefficients whose values are -k and -k', respectively. Values in Table 3-2 are EA, CA, k,
and k' and can be substituted directly into Equations (3-8) and (3-9).
To arrive at the reported fits, any system is removed from the data set if it deviates
by 2.5 times the average deviation, and the fit is rerun. The fit is repeated until all such
systems are removed, with systems previously removed added back if they subsequently
Table 3-1. Ea and CB parameters for nitrogen donors
no. donor wte EB CB CB/EB
1 N(CH3)3 1.0 1.21 5.61 4.6
2 NH(C2H5)2 0.2 1.22 4.54 3.7
3 N(C2Hs)3 1.0 1.32 5.73 4.3
4 NH3 1.0 2.31 2.04 0.9
5 piperidine 0.2 1.44 4.93 3.4
6 pyridine 1.0 1.78 3.54 2.0
7 3-CH3-pyridine 1.0 1.81 3.67 1.9
8 4-CH3-pyridine 1.0 1.83 3.73 2.1
9 NH(CH3)2 1.0 1.80 4.21 2.3
10 NH2(CH3) 1.0 2.16 3.12 1.4
11 quinoline 0.5 2.28 2.89 1.3
12 NH2(C2Hs) 0.7 2.34 3.30 1.4
(a) Weights are those assigned in reference 12a.
deviate by less than 2.5 times the average deviation. Reported donors that are omitted
from the final fits are discussed in the text and listed in Table 3-2.
For adduct formation enthalpies involving a series of bases forming 2:1 adducts,
Equation (3-8) is solved for the acceptor parameters EA, CA, k, and k'. Other
physicochemical properties, AX, are fit to Equation (3-9) where constant energy terms or
intercepts are represented with a W term. Care must be taken to consider all of the
independent energy contributions to a reaction in fitting data to equations and in
interpreting fit parameters. The energy components of each reaction or spectral shift are
described in the results and discussion section.
Some of the systems discussed in this paper were analyzed using a steric onset
term, s(O OoN)8. This term, described above, was added to the multiple regression
analyses by the following procedure derived earlier (Chapter 2). The larger cone angle
systems are removed systematically until a good fit results. A plot of the deviations from
this fit versus 0 gives a line whose slope is s and whose intercept at zero deviation is S0oN.
Results and Discussion
Considerably more information about the influence of electronic, X-back bonding,
and steric effects on physicochemical properties can result when the enthalpies of the
individual steps of 2:1 adducts are determined separately. The first step can be fit to
Equation (3-1) and the second step used to define k and k'. Unfortunately, the chemistry
usually does not permit this step resolution, and the total change for both steps is all that
can be measured. The results from fits to Equation (3-9) of several reported data sets are
given in Table 3-2.
Table 3-2. Acceptor Parameters for 2:1 Adduct Formation
system EA CA W k k' num."
-AH; Ni(DBH)b 2.2 2.6 2.50.6 -4.9 1.9 -0.1 1.5 0.5 0.2 15(3)
-AH; [RhCI(CO)2]z2' -46.8 15.8d 5.0 1.5 d 45.0 9.0 -59.0 22.4 1.3 0.5 11(0)
Vco; RhCI(CO)L2c -163.2 24.3 -4.1 2.3 2215 14 -116.1 34.4 4.3 0.8 11(0)
-AH; Fe(BDA)(CO)3f 34.2 28.5 -8.3 4.6 50.7 22.2 84.8 47.8 -2.3 1.2 16(3)
Vco; Fe(CO)3L2 -70.9 23.9 -6.4 5.6 2016 25 -43.0 50.0 1.4 1.2 16(0)
-AH; Mo(NBD)(CO)4h -50.0 2.6 3.9-0.3 57.5 2.8 -38.82.3 1.4 0.1 11(1)
Vco; Mo(CO)4L2' -6.0 9.8 -8.2 0.9 2082 10 2.8 8.4 -1.6 0.4 11 (1)
-AH; CpRu(COD)CI -72.6 14.7 6.4 1.6 75.3 12.8 -94.8 19.8 2.5 0.5 13(0)
-AH; Cp*Ru(COD)CI k -67.3 22.6 8.1 2.4 62.1 19.7 -91.6 31.0 3.0 0.7 12(0)
"3C; Ni(CO)3L' 8.4 0.4 1.9 0.1 184.1 0.5 38(0)
3C; Ni(CO)2L2t= -4.9 8.5 2.4 1.0 185.7 4.2 -17.5 13.6 0.3 0.2 9(0)
Vco; Ni(CO)3L" -54.7 5.2 -12.4 0.7 2145 5 29(0)
Vco;Ni(CO)2L2 207 72 -26.3 3.4 2042 39 433.8 115.5 5.3 1.3 13(4)
vco; Ir (CO)(CI)L2 p 13.5 47.0 -3.1 13.9 1977 29 20.1 60.8 0.2 1.3 15(2)
log k; Ir(CO)(Cl)L2-MeI(1)q -11.2 4.9 7.6 2.0 -34.3 4.5 -31.6 13.9 1.6 0.4 13(2)
log k; Ir(CO)(Cl)L2-MeI(2)q -7.5 2.6 4.8 1.0 -22.4 4.7 -21.5 7.3 1.0 0.2 8(1)
log k; Ir(CO)(Cl)L2 -H2 (1) -26.7 8.9 3.1 2.3 9.3 4.9 -38.3 10.3 1.0 0.2 13 (0)
log k; Ir(CO)(Cl)L2 -H2 (2)' 68.0 34.9 -4.0 2.0 -47.6 23.3 -83.7 43.6 1.9 0.9 6(0)
log k; Ir(CO)(Cl)L2 -02 (f)' -43.4 5.2 11.3 1.9 -26.6 7.5 -86.1 13.1 2.6 0.4 7(0)
log k; Ir(CO)(Cl)L2 -02 (r)t 1.5 2.6 2.4 0.9 -12.4 3.8 -2.4 6.6 0.2 0.2 7(0)
log K; Ir(CO)(Cl)L2-02 U -44.8 3.6 10.0 2.5 -14.2 5.2 -83.8 9.0 2.4 0.3 7(0)
Vco; Mn2(CO)gL2v -129.9 28.9 6.9 7.7 203235 -185.1 70.8 3.2 1.5 13(1)
v; Mn2(CO)gL2w -6.5 1.3 0.40.4 30.7 1.5 -8.23.2 0.1 0.1 12(1)
Vco; Co2(CO)j6L2_ x -70.6 79 -38.4 14.2 2272 55 88.5 153.6 -5.5 3.1 10(1)
system EA CA W k k' num.a
V; Co2(CO)6L2y -2.4 2.9 -1.2 0.7 35.6 2.6 3.0 6.8 -0.2 0.1 8(1)
log k; Mo(CO)2(PR3)2Br2z -184.1 6.6 18.0 0.9 95.02.3 -243.5 10.1 6.70.3 8 (1)
(a) The total number of donors (N, P, S, As, etc.) studied. In parenthesis is the number of donors excluded to give the
(b) Enthalpy for the addition of 2 ligands to Ni(diacetyl bisbenzoylhydrazone) in benzene. Data from ref. 98 include
protonic donors that may hydrogen bond to the solvent. r2=0.98, F=155.5. Phosphorus donors were omitted because of
(c) Enthalpy for the breaking of the dimer and addition of 2 phosphines to each of the monomers with the release of 2 CO in
CH2C12 at 30C. Data from ref. 99. r2=0.98, F=73.4.
(d) For this system, the EA reported here is actually 2 EA + kEco and CA is 2 CA + k'Cco. See text for discussion of these
(e) CO stretching frequency in RhCl(CO)(PR3)2 in CH2C12. Data from ref. 99. r2=0.99, F=294.7.
(f) Enthalpy for the displacement of BDA by 2 ligands in THF at 50C. BDA=benzylidene acetone. Data from ref. 100.
r2=0.98, F=I 11. AsPh3 was omitted for incomplete complexation and P(i-C3H7T)3 and P(c-hex)3 for steric effects.
(g) CO stretching frequency in Fe(CO)3L2 in THF. Data from ref. 100. =r2--0.98, F=130.
(h) Enthalpy for the displacement of NBD by 2 ligands in THF at 30'C. NBD=norbornadiene. Data from ref 101. r=0.99,
F=864.5. P(OPh)3 was omitted.
(i) CO stretching frequency in Mo(CO)4L2in THF. Data from ref 101. r2--0.98, F=92.5. P(OPh)3 was omitted.
(I) Enthalpy for the displacement of COD by 2 ligands in THF at 30"C. COD=cyclooctadiene. See text for the addition of
a steric onset term, s = -0.83. Data from refs. 102a,b. r2=0.99, F=106.
(k) Enthalpy for the displacement of COD by 2 ligands in THF at 300C. COD=cyclooctadiene. See text for the addition of
a steric onset term, s = -0.91. Data from ref. 103. r2=0.96, F=32.
(1) 13C ofNi(CO)3L in CDCI3 referenced from TMS. Data from refs. 65 and 104. r=0.97, F=643.4. This is a 1:1 system,
and no k or k' values are needed to treat the data.
(m) 13C of Ni(CO)2L2 in CDCI3 referenced from TMS. Data from ref 104. r2=0.99, F=285.3.
(n) CO stretching frequency for Ni(CO)3L in CH2C12. Data from ref 105. r2=0.95, F=248.0. ECW analysis is from Chapter
2. This is a 1:1 system, and no k or k'values are needed to treat the data.
(o) CO stretching frequency in Ni(CO)2L2 in various solvents. Data from ref. 106. r2--0.99, F=144.4. P(OC2H5)3, Pph2(n-
C4H9), AsPh3, and PC3 were omitted.
(p) CO stretching frequency in Ir(CO)L2. Data from ref. 107. r2=0.98, F=109.9. P(i-C3H7)3 and P(c-hex)3 were omitted.
(q) log of the reaction rate for the addition of Mel to Ir(CO)CIL2. (1) was done in benzene solvent with r2=0.94, F=34.0;
and (2) in acetone with r2=0.98, F=42.0. Data from ref. 107. P(i-C3H7)3 was omitted from both fits and P(c-hex)3 was also
omitted in the benzene fit. It was not studied in the acetone experiment. Additionally, P(Ph-p-CF3)3 was
omitted in the acetone fit.
(r) log of the reaction rate for the addition of H2 to Ir(CO)CIL2. (1) was done in toluene solvent with r2--0.94, F=31.2; and
(2) in DMF. Data from ref. 107. r2=0.97, F=9.6.
(s) log of the reaction rate for the addition of 02 to Ir(CO)ClL2. Data from ref. 107. r70.99, F=48.8.
(t) log of the reaction rate for the loss of 02 from Ir(CO)CIL2(02). Data from ref 107. r2=0.98, F=25.8.
(u) log of the ratio of forward and reverse rate constants for 02 reaction with Ir(CO)CIL2(02). Data from ref. 107. r2=0.99,
(v) CO stretching frequency of Mn2(CO)sL2 in CH2Cl2 solution. Data from ref 28. r2=0.96, x =l. 10 (average deviation),
F=45.1. PPh2(C2Hs) was omitted.
(w) Electronic absorption of Mn2(CO)gL2 in CH2CI2 and hydrocarbon solutions. Data from ref 28. r2=0.99, x=0.09,
F= 126.1. P(i-C4H9)3 was omitted.
(x) CO stretching frequency of Co2(CO)6L2 in CH2C12 solution. Data from ref. 28. r2=0.91, x1.84, F=12.6. P(n-C4H9)3
(y) Electronic absorption of Co2(CO)6L2 in CH2CI2 and hydrocarbon solutions. Data from ref 28. 2=0.99, x=0.03, F=76.
P(n-C4H9)3 was omitted.
(z) Flash photolysis studies of Mo(CO)3(PR3)2Br2. Log of the second order rate constant for recombination of CO to
Mo(CO)2(PR3)2Br2 in 1,2 dichloroethane. Data from ref. 78. r2=0.99, F=575.2. PPh(C2H5)2 was omitted.
Selected acceptors will be discussed in detail to illustrate the conclusions that can
be drawn concerning reactivity. Using the values from Table 3-2, the components of each
step toward the total enthalpy or shift can be calculated. The calculation is shown for the
two donors studied whose CB/EB ratios are the most different. This helps to illustrate how
the difference in reactivity of the donors affects the contribution to each step. These
values appear in Table 3-3 for each of the systems discussed below.
Interpretation of Enthalpies of Interaction
The enthalpies for the addition of two donor molecules to diacetyl bisbenzoyl
hydrazino nickel (HI), Ni(DBH), have been measured98 in benzene solvent. Correlation of
the reported data to Equation (3-9) is poor. Eliminating various combinations of donors
can produce good fits but unreasonable signs for the parameters. For example, large
negative EA or CA values are unreasonable for correlations involving a bond formation
enthalpies. An excellent data fit results (r2= 0.98) with reasonable parameters, when
phosphines (that are capable of n-back bonding) are omitted and the enthalpy of displacing
benzene by itself is entered as zero. Addition of benzene forces a W value that is close in
magnitude to the benzene interaction enthalpy. The correlation is illustrated by the
plussigns and solid line in Figure 3-1. The donors are numbered in the figure as in Table
3-1. The three phosphines, P(C2Hs)3, P(n-C3H7)3, P(n-C4H9)3, were omitted from the
correlation and give measured enthalpies that are 4-5 kcal mol"' larger than predicted
because of w-backbond stabilization. Thus, the sum of the n-contributions given by Wr-
Table 3-3. Individual components accounting for the total enthalpy or shift for donors
with different CB/EB ratios
Donor EAEB CACB EAdEB CA CB W Total
___________ _____-AH Ni(DBH) _____
N(CH3)3 2.6 14.0 2.8 -1.4 -4.9 13.1
NH3 5.0 5.1 5.7 3.1 -4.9 14.0
P(C2Hs)3 -13.1 27.7 -8.5 -12.1 45.0 39.0
P(Ph-p-CF3)3 -42.6 7.6 6.3 4.6 45.0 20.9
P(C2Hs)3 -45.7 -22.7 -36.6 -154.2 2215 1956
P(Ph-p-CF3)3 -148.5 -6.2 -52.4 -16.2 2215 1992
P(C2H5)3 9.6 -45.9 2.9 24.4 50.7 41.7
P(Ph-p-CF3)3 31.1 -12.6 -39.1 -7.3 50.7 22.8
P(C2Hs)3 -20.5 -32.1 -16.6 -77.9 2013 1866
P(Ph-p-CF3)3 -66.5 -8.8 -25.7 -12.3 2013 1900
P(C2Hs)3 -14.0 21.6 -10.9 -21.2 57.5 33.0
PPh3 -35.0 11.9 -16.0 -1.1 57.5 17.3
P(C2HS)3 -1.7 -45.3 -1.9 3.6 2082 2037
PPh3 -4.2 -25.0 -5.6 -10.1 2082 2037
P(C2Hs)3 -20.3 35.4 -12.9 -41.4 75.3 35.3
P(Ph-p-CF3)3 -66.1 9.7 12.4 3.9 75.3 23.6
P(C2H5)3 -18.8 44.6 -11.7 -48.1 62.1 27.2
P(Ph-p-CF3)3 -61.2 12.3 14.6 5.3 62.1 20.3
I I '3C Ni(CO)2L2
P(C2Hs)3 -1.4 13.3 0 4.5 185.7 202.1
Donor EAEB CACB EA2dEB CATdCB W Total
P(OC6H5)3 -3.5 4.1 5.3 3.2 185.7 194.9
P(C2Hs)3 3.8 -17.2 2.2 -24.3 1977 1941
P(Ph-p-CF3)3 12.3 -4.7 -4.3 -5.3 1977 1975
log k Ir(CO)(Cl)L2 CH3I in benzene
P(C2Hs)3 -3.13 42.17 -0.69 -5.90 -34.32 -1.88
P(Ph-p-CI)3 -9.18 17.92 11.77 9.24 -34.32 -4.56
log k Ir(CO)(Cl)L2 CH3I in acetone
P(C2H5)3 -2.10 26.91 -0.41 -2.87 -22.44 -0.91
P(Ph-p-Cl)3 -6.16 11.44 8.33 6.06 -22.44 -2.78
log k Ir(CO)(Cl)L2 -H2 in toluene
P(C2Hs)3 -7.50 17.02 -4.50 -14.72 9.30 -0.40
P(Ph-p-CF3)3 -24.39 4.68 7.31 2.28 9.30 -0.82
log k Ir(CO)(Cl)L2 -H2 in DMF
P(Ph-p-OMe)3 42.46 -14.39 10.29 9.44 -47.58 0.22
P(Ph-p-CF3)3 62.32 -6.13 -6.98 -1.81 -47.58 -0.17
Slog k Ir(CO)(CI)L2 -02 pickup
P(C2H5)3 -12.13 62.14 -5.37 -18.53 -26.59 -0.48
P(Ph-p-Cl)3 -35.51 26.41 22.40 11.83 -26.59 -1.46
log k Ir(CO)(CI)L2 -02 release
P(C2Hs)3 0.41 6.53 0.59 0.04 -12.38 -4.81
P(Ph-p-CI)3 1.20 2.78 2.78 1.60 -12.38 -4.02
log K Ir(CO)(CI)L2 02
P(C2H5)3 -12.54 55.61 -5.97 -18.57 -14.21 4.33
P(Ph-p-CI)3 -36.71 23.63 19.62 10.24 -14.21 2.56
9 1110 2
...... 3 12
A8 A 10 2 A5
AS9 AA 12
Total measured enthalpy (U) and the calculated 1:1 adduct formation
enthalpy (A) for Ni(DBH). See text for further discussion.
W and multiplying the ir-parameters by EB and Ca in Equations (3-4)-(3-6) is 4-5 kcal
The EA and CA values for 1:1 adduct formation of 2.20 and 2.50 result. The lack
of donors with a small CW/EB ratio leads to a poor definition of the EA and k values, the
former having a t-value of 0.8 and the latter -0.1, but the covalent parameters are well
defined. Enthalpies for the first step to form the 1:1 adduct can be calculated with EA and
CA, and the contributions of this step to the total enthalpy are indicated by open triangles
in Figure 3-1. These calculated values for the 1:1 enthalpies assume W corresponds to
complete dissociation of benzene in the first step. Except for the phosphines, the
difference between the calculated 1:1 values and the total enthalpies give the enthalpies for
the second step. This result is significant because the enthalpies of the individual steps
cannot be measured directly.
As shown in Figure 3-1, the enthalpies for the first step vary widely. The
difference between the 1:1 points and the solid line give values for forming the 2:1 adduct
from the 1:1 adduct. These enthalpies can be calculated with the EA. and CA/
parameters given by Equations (3-5) and (3-6). One notes in Figure 3-1 that donors with
the largest Ca values have the smallest -AH2 as expected from Equation 11 and a k' value
of 0.49. For the purpose of illustration, the 1:1 adduct with the donor pyridine has EA2d
and CAL values of 2.4 and 0.8, respectively. A significant decrease in the relative
importance of covalency results in the second step as a consequence of coordinating the
covalent donor pyridine in the first step. The k value for the EA 2 parameter is not
significant and has a large error. Using Equations (3-4) and (3-7) and assuming all the
benzene is dissociated in the first step, the acceptor parameters calculate enthalpies for 1:1
adduct formation with pyridine of-7.9 kcal mol' and -7.0 kcal mol"1 for the second step.
Incomplete benzene dissociation in step 1 would increase the enthalpy of step 1 and
decrease step 2.
The decreased covalent contribution in the 2:1 adducts gives rise to a large trans
effect in this complex that is covalent in nature. The trans effect is expected because both
ligands interact with the same set of metal orbitals. Spin state changes also contribute to
the acidity differences of the 1:1 and 2:1 acceptors.
The enthalpies of reaction of [Rh(CO)2C1]2 to form 1:1 adducts with a series of
donors that are not n-acceptors have been correlated earlier' using the ECW model. All
enthalpies correlated in ECW are in terms of kcal mol"' adduct so the quantities fit, -AH,
are one-half the reported enthalpies of reaction. Using the most recent set of donor
parameters,12' the acceptor parameters for 1[Rh(CO)2CI]2 ofEA = 4.32, CA = 4.13, and W
= -10.39, where W is attributed to one-half the enthalpy of dimerization of the acid. The
application of Equation 1 to complexes of the phosphines produces EA*Fr, CA*r and
Wrr parameters that differ from those above because of the 7t-contributions shown in
Equations (2-10) (2-12). Unfortunately, the enthalpy for only one phosphorus 1:1 donor
adduct has been reported.
Reported 2:1 enthalpies involve reaction of [Rh(CO)2ClI2 with a donor (B) to
cleave the dimer and displace one CO forming RhB2(CO)Cl. The enthalpies are measured
in CH2C12 solution where phosphine donors hydrogen bond, and the enthalpy of
dissociation of the hydrogen bonding can be incorporated into the resulting fit
parameters.12' In the second step of the reaction, a molecule of CO is dissociated by the
donor. This enthalpy contribution would lead to Equation (3-10) where EAE",co and
-AHT = EA*FrEB + CA*F CB + EAEB + CA2CB EA2ECo CA2CCO + W (3-10)
and EA*Fr and CA*Irr indicate the acceptor backbonds to the phosphine. Substituting
Equations (3-5) and (3-6) for EA2d and CA2 leads to:
-AHT = 2EA*r EB + 2CA*nT CB kEa2 k'Ca2 EAECo kEBEco -CACco-k'CaCco +W
By incorporating EAECO and CACCO into W, k'CBCco into CA*rrcB, and kEBEco into
EA*FrEB, Equation (3-12) emerges,
-AHT + Corr = EA#EB + CACB kEa2 k'Ca2 +W (3-12)
where EA* = 2EA*rr kEco, CA" = 2CA*F k'Cco. In the data fit, the enthalpy in units of
CA2dCcO give the energy to dissociate CO from the 1:1 adduct, kcal mol' ofadduct is
solved for EA*, CA", k, k' and W. As shown above, these EA" and CA# values are the net of
phosphine binding and CO displacement. EA# and CAO also contain the n-component
shown in Equations (2-10) and (2-11) (e.g., EA' =2EA kEm kEco where EA is 4.32).
The W value also contains the dimer cleavage and CO displacement terms described above
and the 7r-contributions of Equation 6. The above analysis indicates that an ECW analysis
is relevant to this data. Indeed, an excellent data fit to Equation (3-12) results as shown in
Because half of the reported enthalpy is for reaction with each monomer, the
quantity used in the fit for these dimeric systems is half of the reported value from
reference 99. While this is a simplification of the system, the use of this value facilitates