Synergism in metal carboxylate clusters


Material Information

Synergism in metal carboxylate clusters
Physical Description:
ix, 231 leaves : ill. ; 28 cm.
Bilgrien, Carl Joseph, 1959-
Publication Date:


Subjects / Keywords:
Metallic soap   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1986.
Bibliography: leaves 222-230.
Statement of Responsibility:
by Carl Joseph Bilgrien.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000989392
oclc - 17689604
notis - AEW6254
sobekcm - AA00004853_00001
System ID:

Full Text







To Deanna,
who always knew.


No scholar or scientist works alone: each must rely on the

labors of past workers and the assistance of his contemporaries.

I have incurred many debts in this regard.

First and foremost I thank Professor Russell S. Drago for his

support, encouragement and advice. He has generously shared his

insight and perseverance and I am grateful for having had the

opportunity to work with him.

I thank my committee members, Professors Earl Muschlitz,

David Richardson, Harry Sisler and E. Dow Whitney for their


For his continued friendship and enthusiasm I am especially

grateful to Dr. Barry B. Corden.

To all the group members who have shared their time, advice,

expertise, grousing and politics, I am grateful. These include

Kenneth Balkus, Iwona Bresinska, Jeffrey Clark, Richard Cosmano,

Shannon Davis, Peter Doan, Andrew Griffis, Karen Jongeward,

Ernest Stine, Joshua Telser, Keith Weiss and Ngai Wong.

For their ability to build anything I drew I thank the men

in the glass shop, Rudy and Dick. For their sweat and good

humor, I thank Vernon, Chester and Daly of the metal shop. For

not giving up on the calorimeter I thank Russell Pierce. For his

many suggestions and services I thank Dr. Roy King.

Finally, I can never repay the time and sacrifices of my

wife, Deanna Saint Souver. For her unabated encouragement

inspite of all too many hours spent in lab, I will love her




ACKNOWLEDGEMENTS................................... iii

KEY TO ABBREVIATIONS............................... vii

ABSTRACT................... ......... ....... ......... viii

CHAPTER I. GENERAL INFORMATION... .................. 1

CHAPTER II. REACTIVITY OF Cr2(O2CR)4............... 14
A. Introduction............... ............ 14
B. Results and Discussion...................... 18
C. Conclusion ................................. 67
D. Experimental................................ 68

METAL-METAL BOND...................... 78
A. Introduction................................ 78
B. Results and Discussion...................... 81
C. Conclusion .......... ... ............... .... 96

M(III)2M(II)(O2CR)6L SPECIES.......... 98
A. Introduction. ............... .............. 98
B. Results and Discussion ..................... 100
C. Conclusion.................................. 125
D. Experimental............................... 125

CARBOXYLATES.. ....................... 129
A. Introduction................................. 129
B. Results and Discussion...................... 132
C. Conclusion ................................. 185
D. Experimental.............................. 188





REFERENCES.................................. ..... 222

BIOGRAPHICAL SKETCH................................ 231


but butyrate = 02CCH2CH2CH3

hept heptanoate = 02C(CH2)5CH3

hfb heptafluorobutyrate = 02CCF2CF2CF3

OAc acetate = 02CCH3

oct octanoate = 02C(CH2)6CH3

prop propionate = 02CCH2CH3

tfa trifluoroacetate = 02CCF3


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



Carl Joseph Bilgrien

August 1986

Chairman: Professor Russell S. Drago
Major Department: Chemistry

Synthetic, spectroscopic and reactivity studies on several

metal clusters with carboxylate ligands are described. These

complexes are of the general formula M2(02CR)4L2 or M'3(02CR)6L3

where M is Cr or Rh; M' is Co, Cr, Fe or Ru; 02CR is a bridging

carboxylate ligand; and L is a neutral donor ligand. These

studies were undertaken to examine the influence of reaction at

one metal center upon that at an adjacent metal atom and to help

understand the metal-metal bonding interactions which contribute

to the transmission of coordination effects. Ligand exchange

reactions of Cr2(O2CCF3)q[(CH3CH2)20]2 were monitored.

Equilibrium constants and enthalpys for exchange reactions with a

variety of donors were determined from calorimetry data. The

resulting enthalpys were used in a correlation analysis which

demonstrated that the Cr(II) centers are significant Lewis acids

and interact with axial ligands almost exclusively in an


electrostatic sense. Despite a relatively weak metal-metal bond

the first exchange enthalpy is appreciably higher than the

second. Magnetic susceptibility measurements show increasing

paramagnetism for the Cr24+ unit as stronger donors displace

coordinated ether. Strong donors promote oxidation and

rearrangement of the dinuclear unit.

Mixed ligand complexes of the form BRh2(O2CCF2CF2CF3)4L where

B is a Lewis base were examined by electron paramagnetic

spectroscopy (L is the spin label 2,2,6,6,-tetramethylpyridine-N-

oxyl) and infrared spectroscopy (L is CO) and the spectral data

used to calculate acid parameters which describe the Rh2 unit.

The spin label g-value, the CO stretching frequency and the

calorimetric enthalpy, all of which describe the perturbation the

base makes on the trans-metal atom, show different relative

electrostatic/covalent responses as the base is varied,

demonstrating the method dependence of monitoring donor-acceptor

adduct formation. The electronic spectra of clusters of the form

M30(02CR)6L3 (M = Co, Fe, Ru) exhibit a donor ligand (L)

dependence only for M = Ru.

Lastly, clusters of the form Ru30(02CR)6L3n+ (n = 0,1) are

shown to catalyze the selective oxidative dehydrogenation of

alcohols. A mechanistic proposal incorporates the demonstrated

nonradical behavior and the observed stoichiometry. Nonradical

chemistry and reduction of oxygen to water are demonstrated and a

mechanism proposed. High catalyst activity is suggested to arise

from the multiple metal centers acting in concert.



An underlying theme in the current and recent intense

interest in the chemistry of molecules with multiple metal

centers is the way in which the metal centers influence each

other and generate unique reactivity. This synergistic interplay

is implicated in a host of chemical systems. Many enzymes rely

on multiple metal centers for substrate binding (e.g. hemocyanin,

laccase) while others employ proximal metal atoms for electron

transport (e.g. cytochrome c oxidase) or perturbation of the

substrate binding (e.g. nitrogenase). Examples of homogeneous and

heterogeneous catalyses with multiple metal centers abound

whereas interesting physical properties or reactivities of a

stoichiometric nature are often introduced by metal atoms in

close proximity.

The nature and extent of discrete metal-metal interactions

can vary from direct orbital overlap to long range electron

transfer. Studies along this continuum have proceeded via

several fronts: physical and theoretical studies of orbital

interactions; introduction of specific reactivity or binding;

modeling of metalloenzymes; and studies of electron transfer.

One area in particular, the study of molecules with metal-metal

bonds, has received considerable attention following the

pioneering work in the laboratories of Wilkinson1-3 and

Cotton.4-5 The latter area is the focus of this thesis; a

historical perspective is given here as introduction.

Indentification of single and multiple metal-metal bonds as

stereoelectronic elements of molecular structure has come about

only recently. Prejudiced by the conceptual framework of

coordination chemistry established by Alfred Werner6, metal-metal

bonds were deemed inconsistent with one center coordination

chemistry. The advent of x-ray crystallography heralded

recognition of bonding between metal atoms. Although a

crystallographic report7 as early as 1946 demonstrated Mo-Mo bond

distances shorter than those in metallic molybdenum, unequivocal

evidence for metal-metal bond formation came from the structure

determination8 of Mn2(CO)10 in 1957. Evidence for multiple

metal-metal bonding (in this case, a quadruple bond) came from

the initial structure determination9 of the Re2C182- anion and

its reinvestigationI0 in 1964. These early discoveries and the

chemistry of metal-metal bonds have been throughly reviewed

through 1980 in "Multiple Bonds Between Metal Atoms" by F. A.

Cotton and R. A. Walton.11 Several other more specific reviews
12 16
have appeared recently.12-16

The importance of the Re2C144- anion to the development of

this field lies in the electronic structure of the Re-Re bond.

The crystal structure17 of K2(Re2Clg) (H20)2 demonstrated a Re-

Re distance of 2.24 A (2.75 1 in metallic Re) and eclipsed

conformation of the two ReCl4 halves. Both of these features can

be qualitatively explained by considering d orbital overlap. The

in phase and out of phase combination of pairs of d orbitals on

the two metal atoms generates five bonding and the corresponding

antibonding molecular orbitals (MO) in Dch symmetry. The in

phase bonding interactions are shown schematically in Fig 1-1.

Positive overlap of the dz2 orbitals generates a o bond (Fig 1-

la) whereas negative overlap generates the corresponding a

antibonding orbital (not shown in Fig 1-1). The dxz and dyz

combinations give rise to two orthogonal but degenerate Tr bonds

and the corresponding 'r antibonding orbitals. Lastly, in the

absence of any ligand interactions, cofacial overlap of the pairs

of dxy and dx2 _2 results in degenerate 6 bonds and their 6

complements. Introduction of ligand orbitals lowers the symmetry

to D4h and removes the degeneracy of the 6 interactions (the

dx2 _2 orbitals point towards the ligands; the dxy orbitals point

between them). The dx2_y2 orbitals are thus utilized in ligand

bonding and play effectively no role in metal-metal bond

formation. The extent of orbital overlap dictates the MO

energies and the orbital diagram which results is shown in Fig 1-

2. In terms of the (Re2C18)2- anion, each Re3+ center

contributes four electrons to give a net 02S62 quadruple bond,

which explains the short Re-Re distance. The twist angle

dependence of the 6 interaction results in maximum overlap when

the Cl- are eclipsed. Thus, the stabilization gained in 6 bond

formation is greater than the repulsion energy of the eclipsed

halides. The qualitative features of this MO description are

supported by quantitative calculations (SCF-Xa-SW) on the

(Mo2C18)4- ion18-19 and the (Re2C18)2- ion20-21 and is general

for dimers of the 2nd and 3rd row d2-6 transition metals with

octa-halo ligand sets.

'l z

a) C~0 dz CGD




e) 4 dx-y


Fig 1-1. The five nonzero d-d overlaps between two metal atoms.


c) %






Fig 1-2. Qualitative description of the primary metal-metal d
orbital interactions for a dinuclear metal carboxylate in D4h
symmetry. The dx2- 2 orbitals are involved with carboxylate
bridge orbitals and do not contribute significantly to the metal-
metal bond.

The utility of this MO description extends to other ligand

sets as well, most notably the tetracarboxylates, M2(02CR)4L2.

As with the octahalogenates, work with these complexes appeared

in the literature long before their general structural features

were revealed by a crystallographic study22 of Cu2(02CCH3)4(H20)2

(which, incidentally, carries no metal-metal bond) in 1953. The

tetracarboxylate ligand framework is ubiquitous in transition

metal chemistry; examples exist for vanadium, chromium, cobalt,

copper, molydbenum, technetium, ruthenium, rhodium, tungsten,

rhenium, and osmium where L represents a neutral donor molecule

or coordinated anion (naked clusters with no L also abound). The

general structure of M2(02CR)4L2 type dimers is shown in Fig 1-

3. Symmetry dictates the qualitative orbital interactions

displayed in Fig 1-2 apply to the D4h tetracarboxylates also, as

borne out in extensive numerical calculations on dimers of the

2nd and 3rd row transition metals. Dinuclear carboxylates of the

first row, however, are not adequately described by this orbital

description. Representative clusters,23-26 V2(5H5)2(CF3CO2)4,

Co2(quin)2(C6HSCO2)) and Cu2(H20)2(CH3CO2)4 all exhibit longer

metal-metal distances than found in the parent metal. As such,

no metal-metal bond is proposed. For neutral dinuclear

tetracarboxylates of these metals, the observed magnetic behavior

is rationalized in terms of antiferromagnetic exchange between

equivalent spin centers, presumably through the carboxylate

The term dimerr" is traditionally reserved for addition
complexes formed from two monomeric units. In the present
context, the term "dinuclear complex" would be more appropriate;
dimerr" is favored in the interest of brevity.









Fig 1-3. The general structure of M2(O2CR)4(L)2. Diffraction
studies support minor deviation from idealized D4h symmetry.

bridges. The Cr2 center of the chromium tetracarboxylates is

the exception to the rule. Although there has been disagreement

as to the nature of the bonding in the Cr2(O2CR)4L2 systems,

that some bonding interaction is present has been reasonably

established. Results of SCF-Xa-SW27 and ab initio28-29

calculations support a weak quadruple bond as the generic MO

description of Fig 1-2 would suggest for two d4 centers.

As a family, the transition metal tetracarboxylate dimers

comprise a series of stable complexes, structurally well

characterized in most cases, whose electronic structures and many

physical properties can be interpreted in terms of a generic MO

description. As such they provide an excellent opportunity to

study the effect of reactivity at one metal center upon another.

This idea has been exploited by Drago and coworkers in the series

of complexes30-35 Rh2(but)4, Rh2(pfb)4, Mo2(pfb)4, Ru2(but)4C1

and Cu2(hept)4 (but = butyrate, pfb = perfluorobutyrate, hept =

heptanoate). In these studies, stepwise adduct formation of the

Lewis acid dimer with first one, then two donor molecules was

monitored. Lower measured enthalpies for the second reaction

indicated a weakening of the second metal center acidity as a

result of base coordination to the first metal atom. By studying

a range of bases with each dimer, the enthalpy data were treated

in terms of the Drago-Wayland E and C equation.36-39 An

empirical model31-32 was put forward to describe the lowered

acidity in terms of the ability of the metal-metal bond to

transmit electrostatic (E) and covalent (C) effects. The shorter

quadruple Mo-Mo bond was found more able to transmit

electrostatic effects while the longer, more polarizable, single

Rh-Rh bond was found more able to transmit covalent effects.

Consistent with the generic MO scheme in Fig 1-2, the 14 electron

rhodium dimers, with filled T orbitals, undergo TT-back bonding

to IT acids (e.g. CO, pyridine) while the 8 electron Mo2(pfb)4,

with no TT density, does not. The 11 electron Ru2(but)4C1, with

half filled TT orbitals exhibited intermediate interactions.

As mentioned above, description of the metal-metal bond in

the chromium carboxylates has been the focus of recent

controversy. Ab initio calculations40 suggest that correlation

effects are very important in the description of the Cr2

complexes. These calculations involve description of the ground

state wavefunction of Cr2(02CH)4 in terms of contributions from

a2 462 and excited states such as 2 462. The contribution of

o2,462 is 16% which contrasts markedly with similar calculations

for Mo2(02CH)4 in which the C2T'462 configuration contributes 67%.

That is, the quadruple bond adequately describes the ground state

of Mo2(02CH)4. It does not do so for Cr2(02CH)4. deMello et al.

have suggested4142 that the dominant description of the bonding

in Cr24 is one of two Cr atoms antiferromagnetically coupled

with some contribution from multiple bonding. The chromium

carboxylate dimers are also unique in exhibiting a strong

dependence of the metal-metal bond length (and hence electronic

structure) upon axial ligation. Depending upon the nature of L

and R, molecules of the type Cr2(02CR)4L2 display a 0.57 A range

of bond lengths 3- from 1.97 A to 2.54 A. By contrast, adducts

of the Mo and Rh carboxylates display metal-metal bond length

ranges of 0.13 and 0.12 A, respectively. The sensitivity of the

metal-metal bond to donor molecule coordination may be manifested

in the way coordination at one chromium center affects the

second. A quantitative study of the coordination properties of

the Cr2 unit is the subject of chapter two. A literature

report of observed paramagnetism in the chromium dimer studied

here along with the antiferromagnetic description put forward by

deMello et al.42 prompted a magnetic susceptibility investigation

which is also reported in chapter two.

Considerable effort has been expended upon the coordination

chemistry of the carboxylate dimers towards understanding the

nature of the metal-metal bond--its electronic structure,

theoretical description and physical properties and reactivity.

The effects of varying both the axial ligand and the bridging

chelate on the metal-metal length bond have been extensively

explored.11 The perturbation of an axial ligand upon the metal-

metal bond has been termed a trans influence,45-48 referring to

the influence of a ligand upon the bond directly trans to it. In

general, the axial ligand bond competes with the second metal as

a ligand, weakening the metal-metal bond; and conversely, the

stronger the metal-metal bond, the weaker is the metal ligand

interaction.45 Another way of viewing the effects of axial

coordination is the influence of the ligand upon a second ligand

opposite the metal-metal bond. This secondary trans influence or

inductive effect can serve to identify the primary orbital

interactions in the metal ligand bond and the extent to which the

metal-metal bond transmits the ligand influence.

An infrared study of a series of L-Rh2(pfb)4-CO adducts and

an EPR study of a series of L-Rh2(pfb)4-TEMPO (TEMPO is the free

radical 2,2,6,6-tetramethylpiperidine-N-oxyl) adducts has been

performed49 and is further examined in chapter three.

A natural extension of these studies in the transmission of

bonding effects in dimers would be to consider trinuclear

complexes. Retaining the carboxylate ligand set still allows one

to choose from a diverse field of trinuclear complexes (trimers).

A desire to work in noncoordinating solvents and the need for

identical metal sites dictated the use of neutral, trigonal

complexes, exemplified by complexes of the basic iron acetate

structure.50-51 These complexes, of general formula

(M30(RCOO)6L3)n+, where L is a neutral monodentate ligand,

contain a triply bridging oxide ion at the center of a (generally

equilateral) triangular array of metal ions; their structure is

illustrated in Fig 1-4. The electronic and structural details of

many of these compounds with M3n+(n=1) have been studied. The

neutral, mixed valence compounds with n=O have received special

attention as models for intramolecular electron transfer;

examples are known for M = Fe, Cr, Ru, Mn, V and perhaps Co.52"57

A generic MO description for the mixed valence trinuclear

carboxylates is not available, and definitive MO calculations

remain prohibitive in light of their complexity. At least one

system, Ru30(02CCH3)6(PPh3)3, has been addressed from a LCAO

perspective, however.54 Understandably, assignment of the

electronic spectra of the trinuclear carboxylates remains

ambiguous. With regards to the neutral mixed valence trimers,

electronic spectra have been reported only for complexes of Co,

Fe, and Ru.57-59 To gauge the effect of ligand substitution

reactions on their electronic structures, representative neutral

Fig 1-4. General trinuclear, oxo-centered, basic metal
carboxylate structure of formula [M30(RCO2)6L3]n+.

mixed valence trimers were examined in coordinating solvents and

the results are presented in chapter four.

The optical spectra of both neutral and cationic ruthenium

carboxylate trimers exhibit composite bands which originate from

a series of closely spaced molecular electronic transitions.59

The cluster system Ru30(0Ac)6(py)33+/2+/+/0/- displays an

extensive reversible redox chemistry,60 prompting the name

"electron sponge". The stability of different redox states and

availability of substrate binding sites in the ruthenium trimers

are promising for homogeneous redox catalysis. The triruthenium

acetate clusters have shown utility as homogeneous hydrogenation

catalysts for unsaturated substances.61-62 Attempts to employ

these same clusters as olefin oxidation catalysts in this

laboratory revealed the reversible reduction by alcohol solvent

at elevated temperatures. Subsequent specific catalytic

oxidative dehydrogenation of a range of alcohols and mechanistic

features were explored. These findings are given in chapter




A. Introduction

The dimeric metal carboxylates, M2(02CR)4, are convenient

clusters for studying the effects of coordination at adjacent

metal sites. An extensive array of complexes has been isolated

and structurally characterized, synthesis methodology is

relatively straightforward, charge neutrality allows study in

non- or weakly coordinating solvents, and the metal centers

display open axial (trans to the metal-metal bond) coordination

sites to which Lewis bases readily bind.11'32 Work in this

research group has focused upon quantitative description of the

Lewis acid centers. It has been found that the enthalpy measured

for formation of the second metal-base bond is less than that for

the first metal-base bond in dimers of rhodium and molybdenum,

indicating that this may be a general phenomenon of this family

of compounds. By working with a range of characterized bases,

the enthalpy data could be treated with the Drago E and C

model.36-38 The empirical equation (2-1)

AH + W = EAEB + CACB (2-1)

describes the enthalpy of adduct formation where EA and CA are

the acid parameters, and EB and CB are the base parameters

corresponding to the tendencies of the acid or base to undergo

electrostatic or covalent interactions. The W term is included

when any constant contribution to the measured enthalpies

independent of acid or base variation accompanies adduct

formation. Drago, Long and Cosmano suggested an inductive

transfer model to describe the Lewis acidity of the second metal

center.31,32 In this model, the acid parameters of the 1:1

adduct are reduced from that of the free acid by an amount that

is proportional to the corresponding base parameter.

EA1:1 = EA kEB (2-2)

1:1 = CA k'CB (2-3)

The k and k' have physical significance and represent the ability

of the metal-metal bond to transmit electrostatic and covalent

coordination effects. This description can be thought of as

parameterization of a trans effect. To date, four dinuclear

metal carboxylates have been studied and are summarized in Table

2-1. It should be emphasized that the intent of this methodology

lies not in determination of EA and CA numbers per se (although

the experimentally determined numbers can be used to predict

unmeasured enthalpies), but their relative magnitudes serve to

illustrate the nature of the metal-ligand and metal-metal bonds.

For example, both Rh2(pfb)4 and Mo2(pfb)4 exist in the +2

oxidation state and contain the same bridging ligand; a similar

Table 2-1.

Acidity Parameters of various metal carboxylate






MTn+ dn BO

A A. A

4 14 1 3.21 1.32 0.411

4 14 1 5.06 1.74 0.344

4 8 4 5.92 0.385 0.065

5 11 2.5 7.73 1.27 1.64

k k'

1.16 0.0364

a -

1.46 0.022

a.) Not determined though the butyrate and perfluorobutyrate
bridges were found similar in their transmission capability.

b.) Cl1 coordination precludes bonding of a second base.

partial positive charge exists at each metal center, and the two

dimers have similar EA numbers. The less electronegative bridge

in Rh2(but)4 results in a lower EA for this complex. The CA

numbers, on the other hand, reflect the polarizability of the

metal-metal bond. The quadruple metal-metal bond of Mo2(pfb)4 is

not as likely to redistribute electron density over the entire

molecule as the more flexible, single Rh-Rh bonds. The molydenum

carboxylate has a lower CA.

Similar rationale lends physical significance to the

transmission coefficients, k and k'. The shorter metal-metal

bond in Mo2(pfb)4 allows for greater electrostatic interaction of

the second molybdenum with the first coordinated base and greater

interaction of the two base molecule dipoles. The greater

polarizability of the metal-metal bond in Rh2(pfb)4 allows for

better electron density redistribution as manifested by the

larger k' value.

The analogous chromium carboxylates have such a strong

tendency to coordinate electron pair donors in the axial

positions that they are only rarely seen without ligands. In

the two cases where unsolvated chromium dimers were studied

structurally,44,63 axial coordination occurred by association of

the molecules to form infinite chains. The nature of the axial

ligand has a pronounced effect upon the Cr-Cr distances446465

which range from 2.214 X in the (Cr2(CO3)4(H20)2)4- ion to 2.541

Sin Cr2(02CCF3)4(Et20)2. The experimentally observed range of

bond distances would suggest a shallow potential well for the

Cr2 unit. The marked dependence of the electronic structure of

the Cr2 unit upon the axial ligand provides a unique

opportunity to study transmission of bonding effects.

Reactivity studies of the dichromium tetracarboxylates have

focused primarily on the acetate which has found wide utility as

a reducing agent and as a starting material in the preparation of

other compounds containing the Cr2 unit. To this end, dimers

have been isolated and structurally characterized for a variety

of bridging ligands with C, N and 0 donor atoms and a range of

axial bases. Comparison of the structural parameters shows no

clear relationship between the nature of the bridging ligand and

metal-metal bond strength/length. Axial ligation, however, is

found to strongly influence the Cr-Cr bond with stronger donors

generally dictating longer bonds.11

Unlike the strongly bonded Mo24+ ion, there is no evidence

for the existence of the naked cluster Cr2 Bridging ligands

may play a role in metal-metal bond formation other than keeping
the metal centers in close proximity; the (Cr2(CH3)8) and

(Cr2(CHg8)4)4- ions exist without bridging ligands.66'67

The Cr2(tfa)4(Et20)2 adduct was first reported in 1966 to

exhibit weak paramagnetism. The structural report which followed

showed the dimer to contain the longest Cr-Cr bond known. Our

interest lay in probing the transmission of bonding effects

across such a weak, loosely interacting metal-metal bond, and

reactivity studies are reported here.

B. Results and Discussion

1. Qualitative Reactivity

Initial investigations were performed with the simple

carboxylates Cr2(OAc)4(H20)2 and Cr2(but)4(H20)2. The hydrates

are easily desolvated by heating in vacuo. An x-ray diffraction

study63 of anhydrous Cr2(OAc)4 prepared by sublimation of the

hydrate demonstrated that bridging oxygens of neighboring

clusters satisfy the strong coordination requirements of the

Cr(II) centers. A portion of the polymeric compound which

results is shown in Fig 2-1; this compound is soluble only in

coordinating solvents. To minimize the nucleophilicity of the

carboxylate oxygens, the trifluoroacetate bridged dimer

Cr2(tfa)4(Et20)2 was utilized. Again, the need for a

coordinating ligand attests to the Lewis acidity of the metal

atoms. Here, however, axial coordination is superceded by a weak

donor. Adduct formation in these studies proceeds via an

exchange reaction to displace diethylether.

Weak donors such as acetonitrile do not displace ether.

Intermediate donors such as dimethylacetamide cleanly displace

coordinated ether to give first 1:1, then 2:1, adduct formation.

Strong donors such as pyridine rapidly cleave the complex videe

infra). A good measure of solvent donor capabilities come from

their E and C numbers; donors studied are summarized in Table 2-


For the intermediate case, donor exchange is an equilibrium

process, readily monitored by changes in the electronic spectra

of the Cr2 chromophore. Analagous to earlier studies68 with

Rh2(pfb)4 and Rh2(but)4, evidence for the formation of 1:1 and

2:1 adducts of Cr2(tfa)4 is provided by spectral studies.

Representative spectra of Cr2(tfa)4(Et20)2 and its adduct

exchange forms with DMA are shown in Fig 2-2 where the


c / I/ /
O O- Cr Cr,-
\ I/ I/ /I /I N
0-Cr-Cr- O\ 0.
/ /I /I C
Cr --0 0 R

Fig 2-1. The formation of infinite chains of Cr2(02CR)4
molecules by oxygen bridge bonding.

Table 2-2. Donor parameters. E C

diethyl ether (Et2O) .963 3.25

do not displace Et20

acetone .987 2.33

acetonitrile .886 1.34

dimethyl sulfide .343 7.46

methyl acetate .903 1.61

tetrahydrothiophene .341 7.90

triphenylphosphine (a)

displace Et20

dimethyl acetamide (DMA) 1.32 2.58

dimethyl cyanamide (DMCA) 1.10 1.81

dimethyl formamide (DMF) 1.23 2.48

dimethyl sulfoxide (DMSO) 1.34 2.85

dimethyl thioformamide (DMTF) (a)

dioctyl ether (DOE) 1.10 3.40

p-dioxane 1.09 2.38

hexamethylphosphoramide (HMPA) 1.52 3.55

triethylphosphate (1.36)b (1.81)

trimethylphosphine oxide (1.53)b (3.32)

trimethylphosphite (1.03)b (5.99)

tetrahydrofuran (THF) .978 4.27

tetramethylurea 1.20 3.10

dissociate complex
diethylamine 1.17 8.51
N-methyl imidazole .934 8.96
pyridine 1.12 6.89
quinuclidine .704 13.2

a) unknown

b) tentative parameters calculated from limited data sets





0 4.)
o 4 r4


0 S ao
0 4- 0

0 .
43 4-) -0

+ 0 Lt O
0 97e
0 0 4)
4 S r-

S. IS.. 0
4) CO.4 e





concentration of the dimer is kept constant and the donor

concentration gradually increased.

The presence of an isosbestic point at 572 nm is evidence for

only two species in solution at low base concentration: the free

acid, A, and the 1:1 adduct, AB. These spectral curves define

the equilibrium in Eqn 2-4.

A + B -i AB (2-4)

Further base addition results in spectral deviation from the

first isosbestic point as a third species is formed in solution,

Eqn 2-5.

AB + B Z AB2 (2-5)

In these equilibria, the free acid, A, refers to the bis ether

adduct, Cr2(tfa)4(Et20)2. Dissociation of coordinated Et20

accompanies both equilibria. At high base concentration, an

isosbestic point appears at 515 nm which upon cursory

examination would appear to correspond to the second equilibrium

process in Eqn 2-5. Quantitative analysis (vida infra), however,

reveals that the limiting spectrum centered at 585 nm

corresponds to the AB2 chromophore while the isosbestic point at

575 nm can be assigned to yet another equilibrium, Eqn 2-6.

AB2 + B -- AB3 (2-6)

The absence of a clearly defined isosbestic point for the second

equilibrium implies one of two things: 1) Kl>> K2, butthe

spectral change associated with Eqn 2-5 is slight, and an

isosbestic point cannot readily be discerned. 2) Ki is not

significantly greater than K2, and no appreciable amount of AB

forms in solution. The presence of a well defined isosbestic

point at low base concentration argues against the latter while

quantitative results videe infra) indicate the former.

The observed spectral changes are understandable in terms of

the primary orbital interactions. Rice et al.69 have examined

the single crystal polarized electronic spectrum of red chromous

acetate dihydrate, Cr2(02CCH3)4(H20)2 whose spectral features are

similar to that of Cr2(tfa)4(Et20)2. Two bands are observed for

the former; the lower energy (465 nm, e = 120 M-1 cm- )

transition is associated with a metal centered 6 ->TT promotion;

the other (333 nm, e = 200) is attributed to charge transfer

from a nonbonding carboxylate T orbital to metal centered T

(n- ->1 ). Violet Cr2(tfa)4(Et20)2 displays a similar spectrum.

The 6 ->Tr* (550 nm, e= 133) and n -> r* (328 nm, = 380)
transitions are assigned by analogy. An MO diagram of the metal

centered orbitals for Cr2(tfa)4 and their changes upon adduct

formation are shown in Fig 2-3. The dimer functions as a Lewis

acid, accepting electron density in the antibonding a orbital.

Upon complexation of a stronger donor, the metal a and a

orbitals become closer in energy while adduct formation is

realized through stabilization of the donor lone pair orbital.

The dimer orbitals (6-> ) involved in the transition are not

directly involved in adduct formation. Weakening of the metal-

metal bond through partial population of a decreases the d



i ---

b t



orbital overlap, compressing the entire d orbital manifold in the

MO scheme. Replacing coordinated ether with stronger donors

gives a color change from violet to blue, consistent with the

expected red shift from the MO description.

Displacement of coordinated Et20 by stronger donors was

monitored by FTIR in order to verify the exchange processes.

Within the detection limits of the FTIR, coordinated Et20 was

completely exchanged for coordinated donor at 2:1 donor:dimer

molar ratio, indicating the exchange equilibria (Eqns 2-4 and

2-5) lie far to the right. Figure 2-4 shows an FTIR titration of

Cr2(tfa)4(Et20)2 with 0, 0.5, 1.0, 2.0, and 5.0 equivalents of

dimethylacetamide. Only above 2 equivalents is free DMA clearly

discernible. Figure 2-5 shows the ether v-o-c region of the

same titration. Bound ether appears completely exchanged at 2

equivalents added donor. Some dissociation of diethylether may

occur even at 0 equivalents added donor as evidenced by a slight

shoulder absorbance at 1113 cm- Only a slight frequency shift

is observed for \v of coordinated dimethylacetamide as the

titration proceeds. A similar result is observed for the

asymmetric vo-c-o stretch of bridging trifluoracetate, vco2,asy:

equiv. DMA vco(DMA) co2,asy

0 1680.1 cm-1

0.5 1610.0 cm-1 1680.3

1.0 1608.2 1682.1

2.0 1606.2 1684.0

5.0 1606.2 1683.9

free DMA


\AAA 0





1700.0 1600.0 1500.0

Fig 2-4. FTIR titration of Cr (tfa)4(Et20) (8.9 X 10-3M in
CH2Cl ) with 0, 0.5, 1.0, %.0 and 5.0 equivalents of
dimethyl cetamide. Bound DMA: v o = 1160 cm-1; free: v
1639 cm .





1150.0 1100 10500 1000.0

Fig 2-5. FTIR titrationn of Cr2(tfa)4(Et20)2 (8.9 X 10-3 M in
CH2C1l) with 0, 0.5, 1.0, 2.0 and 5.0 e uivalentsof
dimethylacetamide. Bound Et20: v co = 1053 cm- ; free: vcoc
= 1113 cm-.

Coordinated DMA is relatively insensitive to the nature of the

trans coordinated ligand (Et20 or DMA). The slight shift which

occurs would indicate that DMA coordination is strengthened as a

stronger metal-ligand bond is formed on the opposite side. Such

a "trans strengthening" is surprising in light of results for the

Rh2(pfb)4 dimer (chapter 3). Across the rhodium-rhodium bond, a

sigma donor on one side decreases the strength of a metal-ligand

sigma bond at the other side, as the two donors compete for the

same rhodium d orbitals.

The observed shifts for coordinated DMA are too slight to

resolve into their individual contributions from 1:1 and 2:1

adduct species. The FTIR experiment does provide the satisfying

result that the exchange process is indeed occurring and that it

is essentially complete after two equivalents of donor has been

added at these chromium concentrations.

The trifluoroacetates are convenient spectroscopic labels and

display strong sharp absorbances for the asymmetric and symmetric

carboxylate stretches which have shown utility in differentiating

between unidentate, ionic, bidentate, or bridging coordination.70

The analagous Mo2(tfa)4 displays the symmetric and asymmetric

stretches at 1592 and 1459 cm" (bridging). Upon adduct

formation, bulky phosphine donors occupy equatorial positions,71

and the resulting unidentate assymmetric stretch occurs at ca.

1680 cm -1. The IR spectrum of Cr2(tfa)4(Et20)2 (Fig 2-6) in

methylene chloride is assigned the bridging asymmetric and

symmetric stretches at 1680.1 and 1480.0 cm-1. The similarity

of the molybdenum unidentate stretch (1680) and chromium













o m
0 D
o z


bridging stretch suggests the possibility of unidentate

coordination for the latter in methylene chloride solution.

Observation of only one asymmetric stretch (four unidentate

carboxylates would disociate the complex) and an identical Nujol

mull spectrum rule out this possibility. Though not detected at

five equivalents, a large excess (50 equiv.) of DMA results in an

FTIR spectrum in which another asymmetric carboxylate stretch

appears at 1717 cm-. This, presumably, is the unidentate

carboxylate which arises from equatorial coordination of DMA and

is consistent with the formation of the AB3 species postulated by

Eqn 2-6.

Similar titrations were performed and spectra recorded for a

range of donors. Results for methylene chloride solutions of

Cr2(tfa)4(Et20)2 (ca. 5x10-2M) with two equivalents donor are

given in Table 2-3. Reported are the absorbances for the

appropriate functional groups of the free and completed donor

molecules. Exchange adducts were readily isolated for several

donors by stoichiometric admixture and recrystallization from

benzene, and in all cases except that for DMTF, gave identical IR

spectra to those prepared from the same donor in situ. For those

donors whose functional groups showed a pronounced frequency

shift upon coordination, the FTIR spectra indicate the magnitude

of K1 and K2. In all cases, no free donor could be detected at

the 1:1 level. With two added equivalents, the absorbance

spectra peak areas demonstrate 90% or greater complexation for

DMA, Et3P04, DMCA, DMSO, HMPA, and DMTF. The extent of exchange

could not be readily gauged for the remaining donors. A 1:1

equilibrium process (AB + B-=- AB2) with initial concentrations

Table 2-3. FTIR data for trifluoroacetate bridges and donor

functional groups.










co2, asy
1 2

-- 1664

1680 1592

- 1680

- 1682

1717 1684

- 1682

- 1681

- 1680,


- 1681,


group free completed













0-C-0 -- 1055

C = 0 1651 1606

P-0-C 1034,1037,

979 984

C = N 2217 2239

S = 0 1057 1012,


P = 0 1179 1134

(HMPA)2 1685 1473 P = 0 980 992

(DMTF)2 1679 1478 (d) 1538 1563

(DMF)2 -- 1680 1477 C = 0 1676 1562

[(MeO)3P]2 1681 1477 (e) --

(Me4Urea)2 1685 1477 C = 0 1640 1581

a) reference 70.

b) reference 71.

c) Slow oxidation follows complexation as evidenced by shift in
asymmetric stretch and color change from violet to green over 24
hour period.

d) assignment unknown.

e) Ligand vibrations shifted but specific shifts not assigned.

of 5x10-2 going to 90% completion yields K2 = 1800. Finally, as

with DMA, little or no frequency shift was observed for the bound

donor as the trans ligand was exchanged from diethylether to the

donor of interest.

An attempt was made to gauge the magnitude of the equilibrium

constants, K1 and K2, by calculating the mole fractions of bound

and coordinated diethylether from the fast exchange region 1H

NMR. The 100 MHz FT-NMR spectra of Cr2(tfa)4(Et20)2 and exactly

one equivalent added DMA and HMPA give the following diethylether

shifts in CgDg:

6(-CH-) 5(-CH)
Et20 3.34-q) 1.07t)

Cr2(tfa)4(Et2O)2 5.22 1.40

+ 1 DMA 4.51 1.26

+ 1 HMPA 4.50 1.26

The chromium containing solutions all exhibit broadened

resonances, devoid of any spin-spin splitting. Treating the

coalesced resonance shifts as a mole fraction weighted average of

the free and bound species gives the same results for both HMPA

and DMA. Using the methylene resonances: 64% completed, 36%

free; methyl, 58% completed, 42% free. That different mole

fractions are calculated dependent on the resonance used is a

feature of the paramagnetic complex. Stronger donors raise the

paramagnetism of the Cr2+ center videe infra). The 1:1 adducts

then will display downfield shifted coalescence peaks.

Calculations thus favor a higher concentration of completed ether

than is actually present. This effect should be more pronounced

for the methylene protons which experience a greater contact

shift. The only information these NMR spectra add to the

picture, then, is support for the paramagnetic nature of the

dinuclear metal center.

2. Quantitative Reactivity

Stepwise adduct formation between a metal center and first

one, then two donor ligands is described by two successive

equilibria (Eqns 2-4, 2-5). For the case where formation of the

first bond exerts no influence upon the second, AH1 = AH2 and

entropy considerations predict K1 = 4K2. Inductive effects, as

in the case of the metal carboxylate dimers, perturb the second

equilibrium such that A Hi > AH2 and K1 > 4K2. In this study,

adductformation is accompanied by a dissociation step,and the

reactions are described by exchange equilibria:

AL2 + B 4= ABL + L (2-7)

ABL + B c AB2 + L (2-8)

In subsequent discussion, the AL2 species is referred to as the

free acid and denoted by 0, the ABL as the 1:1 species (1) and

the AB2 as the 2:1 species (2).

a. Electronic spectra

UV-VIS titrations were performed by successive microliter

injections of a concentrated base solution into an inert

atmosphere cuvette containing the dimer solution. Despite the

presence of an isosbestic point for the initial spectral curves,

absorbance changes are too slight to allow satisfactory

definition of K1. In plots of K-1 vs. Ae, the resulting curves

intersect in a region of very small K-1 and small

absorbance/concentration errors can vary K by several orders of

magnitude. Consistent results are obtained, however, for Ac and

these values provide the initial estimates for the E at various


Treatment of all the spectral curves using the program SPEC

allows the computer to calculate the best fit values of K1, K2,

and C0, E and E2 for the wavelengths analyzed (up to five).

The success of the fit relys upon good initial estimates of each

of these parameters. The E0 value comes from the free acid

spectrum. The e 1 values can be calculated at any wavelength

from any curve that passes through the first isosbestic point,

once As has been determined from a K-1 vs. As plot at one

wavelength. The 2 values are estimated from the limiting

spectra which result from just above 2 added molar equivalents of

base. Specific details and precautions have been described.68

The spectral absorbance data are listed in Appendix I.

Iteratively varying some of the spectral parameters while

fixing the remainder allows the computer to uniquely define the

extinction coefficients and the picture which emerges in the same

absorption profile for the free acid, 1:1 and 2:1 species,

successively red-shifted. This is the expected behavior based

upon the MO arguments described earlier. Based upon the large

equilibrium constants videe infra) these three species can be

observed in solution at the appropriate stoichiometry. The

UV-VIS spectra of the dimers with 0, 1, and 2 equivalents of

dimethylacetamide are shown in Fig 2-7, roughly illustrating the

spectral curves which the e 's define. Calculated values for the

equilibrium constants depend upon the initial guesses.

Consistently, allowing the computer to vary only K2 gives values

of 4K2 slightly less (ca. 50%) than K1. Again, however,

absorbance changes are too slight to define both equilibrium

constants, and a wide range of K1, K2 pairs satisfy the spectral

data with impossibly small standard deviations. If K1 were

located, K2 could be uniquely defined, but the lack of a good

estimate for K1 prohibits their determination from the titration

curves. The spectral titrations indicate a relationship between

the equilibrium constants which would be consistent with weak

communication between the metal centers but do not adequately

define the magnitudes of the K's. The utility of these

experiments, then, lies in verifying that the exchange reactions

(Eqns 2-7 and 2-8) do occur, that their equilibria lie far to

the right and that the metal centers do not influence each others

coordination chemistry significantly.

. Calorimetry

As Long has indicated,68 it is best to determine the

equilibrium constants which apportion the experimental heats and

the molar enthalpies from separate experiments, since the four

parameters, K1, K2, A H1:1, and A H2:1 are frequently highly

correlated. When this happens, a situation arises such as that

for the chromium spectral titrations above which a specific

solution relies on definition of (usually) one parameter.


I %
I \ \ \

I \\

400 560 600 700

Fig 2-7. Electronic spectrum of the chromium dimer (methylene
chloride solution) with 0, 1, and 2 equivalents DMA. Given the
large equilibrium constants for donor exchange, these curves
approximate the 0, 1 and 2 species.

Exceptions to the rule are instances in which the first or both

equilibrium constants are very large. Long showed that the

enthalpies for the rhodium systems, Rh2(but)4 and Rh2(pfb)4,

could be reasonably determined from exclusively calorimetric data

provided that K1 was of the order of 108 or greater, as verified

by comparison of the enthalpies determined when the equilibrium

constants were solved for independently. A similar situation

occurs for the chromium calorimetric data, and the enthalpies are

extricable given the large equilibrium constants. Using the

program HEAT, K1 values were fixed and the K2 varied to minimize

the conditional standard deviations associated with the

enthalpies. K1 was then varied by a factor of ten and the

minimization repeated. In this manner, the best K1, K2 pair

which minimizes the enthalpy deviationns was used to define

AH1:1 and AH2:1. The raw calorimetric data and best fit solution

values of each of the parameters are given in Appendix I. It is

not possible to assign deviations to the equilibrium constants

determined in this manner, and it was found that varying K1's of

this magnitude by a factor of 102 had little effect (<1%) on the

enthalpies. Thus, though the enthalpies are uniquely determined,

K1 is simply described as "large" and K2 represents the best fit

value. No inferences can be made for the relationship between KI

and K2 determined in this manner. Table 2-4 contains the results

of the calorimetric titrations.

Several aspects of the trends and some anamolies of the data

set deserve comment. First and foremost, the enthalpies of the

2nd exchange, AH2:1, are consistently lower than those for the

first, AH1:1, in all cases except that for Me3PO. The lowering

Table 2-4. Thermodynamic data for the exchange reaction of
Cr2(tfa)4(Et20)2 with various donors.

Base K1 K2




(MeO) Pb















- AH1:1
(kcal mol-")a



















-A H2:1

(kcal mol1)



















a. Values in parentheses are conditional

b. Enthalpies are tentative; see text.


is slight but measurable, manifesting the inductive effect of the

first coordinated donor upon the second exchange reaction. As

expected, the inductive influence is small, in keeping with the

long, weak metal-metal bond in Cr2(tfa)4(Et20)2, and is consistent

with the K1, K2 relationship inferred by the spectral data videe

supra). Second, the trends in the enthalpy values is in keeping

with our intuitive knowledge of donor strengths, an aspect which

is taken advantage of in quantitative correlations to follow.

The higher AH2:1 >AH1:1 value for Me3PO is inconsistent

with the remainder of the data set and probably results from

subsequent further reactions. The spectral titration for this

system was well behaved only to a 2.0 molar ratio of Me3PO to

dimer at which point the solution began to turn green, indicating

oxidation of Cr(II). Heat evolution during calorimetric

titration for the other donors was essentially complete following

addition of two equivalents of donor. Heat continued to be

evolved in the Me3PO calorimetric titration which was carried out

to a 2.5 molar ratio. The stability of the dimer adducts toward

oxidation is decreased by donor complexation videe infra) and the

Cr2(tfa)4(Me3PO)2 adduct may be readily oxidized via oxygen atom

transfer from the phosphine oxide.

Another inconsistency in the data set is the apparent

anomalously low heats observed for (MeO)3P despite the agreeable

AH1:1 > AH2:1 relation. Again, the spectral titration lends

insight. During intermediate stages of the titration (ca. 1:1)

greatly decreased (< 50%) absorbances were recorded in spite of

the presence of an isosbestic point at lower ratios. This

behavior would arise if 1) extinction coefficients for the 1:1

species were much lower than for the 0:1 or 2:1 species or 2)

solid formation is occurring in this region. The latter was

observed to be a common phenomenon in titrations with donors

containing two potential donor sites. Maintaining a strict 1:1

stoichiometry allowed preparation of polymeric species and

clearing of the solution for reactions with p-dioxane,

tetramethyl urea and 1-methyl imidazole. Solid formation was not

obvious with (MeO)3P but could occur if both phosphorous and

oxygen donor sites are stronger Lewis bases than the ether

oxygen. A low value for AH1, for endothermic desolvation,

however, would be compensated by a high value of AH2 for

exothermic solvation, and this is not observed in the

calorimetric data. The low heats for (MeO)3P remain a surprise

and should be considered tentative.

Solution studies were conducted in methylene chloride solvent

which behaves as a Lewis acid towards donors. Solvation effects

are anticipated and can be corrected for since methylene chloride

has been shown to undergo primarily specific reaction (electron

pair sharing of a Lewis base with a Lewis acid) with donor

molecules.72 To assess their contribution to the experimental

heats, the enthalpy components are specified:

(Et20)Cr2-(Et20) + B-S 7- (Et20)Cr2-B + Et20-S (2-9)

(Et20)-Cr2B + B-S <= B-Cr2B + Et20-S (2-10)

AH1:1 = AH4 + AH3 AH2 AH1 (2-11)

AH2:1 = AH4 + AH6 AH2 AH5


Assuming nonspecific solvation differences between all the

various chromium adduct species to be negligible, the solvent

contributions, AH2 and AH4, can be calculated from the EB

and Cg values for the various donors and the E'A and C'A values

for methylene chloride, 1.66 and 0.01, determined by Drago et

al.72 The primes indicate that these are the best fit values for

methylene chloride adduct formation with a series of donors and

may contain small contributions from nonspecific solvation.

Solvation corrected enthalpies are given in Table 2-5. In three

instances, the donor EB and CB have been determined from a

limited data set and are not well defined. The corrected heats

for (MeO)3P, Et3PO4 and Me3PO are thus considered tentative. To

alleviate propagation of their uncertainties, these three heats

will not be used in the following analysis.

Included in Table 2-5 are the experimental frequency shifts

from the spectral titrations. The observed red shifts indicate

the perturbation on the transition orbitals which results upon

donor exchange and provide another measure of the strength of the

interaction. The perturbations roughly follow the measured

enthalpies and include two donors not looked at by calorimetry.

Dimethyl sulfide titrations gave very small absorbance changes

and heats could not be accurately measured. Spectral changes are

apparent only at high donor concentration in which entropy

effects predominate and the weaker donor displaces ether to give

a blue shift. Me4Urea forms a polymer with the dimer at 1:1

molar ratios and redissolves as more donor is added, prohibiting

calorimetric determination of the enthalpies.

Table 2-5. Experimental frequency shifts and solvation corrected
enthalpies for the exchange reaction of Cr2(tfa)4(Et20)2 with
various donors.


(kcal mol"1)













- A H2:1

(kcal mol')























a. Nature of reaction prohibits enthalpy determination by

b. EB, CB values unknown.

c. Calculated from tentative EB, CB numbers.














c. ECW model

We are now in a position to correlate the solvation corrected

enthalpies (corrected for AH2 and AH4) whose contributions are

given below:

AH1:1 = AH3 AH (solvent corrected) (2-13)

AH2:1 = AH6 AH5 (solvent corrected) (2-14)

For the observed 1:1 heats, AH1:1, with a series of

different donors, the contribution A H1 is independent of the

donor employed and must be treated as a constant, W. Using Eqn

2-1, where AH = H3 and W =AH1, gives the equation:

-AH1: -AH3 + AH1 = EA1:1 EB + CA1:1 CB (2-15)

Using the experimental solvent corrected heats for the six donors

whose EB and CB are well defined along with the EB and CB numbers

from Table 2-2 allows calculation of the constant contribution,

AH1,and the EA1:1 and CA 1:1 associated with the acid which

defines AH3. This acid is the 1:1 adduct (Et20)Cr2(tfa)4. A

plot of the simultaneous equations is shown in Fig 2-8. The best

fit values for EA, CA and W along with their standard deviations

are E A 1:1 = 13.6 (0.75), CA1:1 = -1.57 (0.24) and W = 8.00

(0.94), all in kcal mol -1. The experimental enthalpies used in

the correlation along with those calculated from Eqn 2-15 are

given in Table 2-6 and demonstrate the quality of the fit.

Fig 2-8. Plot of EA vs CA for Cr2(tfa)4(Et20).

CA -

Table 2-6. Enthalpy data used to determine acid parameters for

Usedto determine

EA1:1 and CA1:1

Used to determine

k and k'

- AH1:1,exp

(kcal mol-1)a


3.5 (0.1)

4.1 (0.3)

4.7 (0.2)

5.0 (0.3)

5.0 (0.2)


(kcal mol-1)b



(kcal/mol)c (kcal/mol)d

1.1 (0.3)

1.5 (0.6)


1.7 (0.5) 1.7

1.6 (0.5) 1.7

0.5 (0.4)

a. Solvation corrected values.
conditional standard deviation

Quantity in parentheses is the
for the measured uncorrected

b. Calculated from equation 2-15.

c. Difference between the first and second experimental
enthalpies. AAH = AH2.1 A HI.1 Quantity in parentheses is
the propagated conditional standard deviation, calculated from
a = A + B

d. Calculated from equation 2-17.






Several points about these best fit parameters warrant

comment. The calculated W = AH1 = 8.0 kcal mol 1 refers to

breaking of one of the adduct bonds in (Et20)Cr2(tfa)4(Et20).

The relatively large EA 1: and small CA1:1 demonstrate a

pronounced tendency for this Cr(II) acid center to interact

primarily in an electrostatic sense as might be expected for a

first row transition metal ion. The negative CA1:1 value does

not imply that the dimer interacts covalently in an antibonding

sense but simply that this is the best fit parameter to a data

set in which implicit assumptions have been made in defining the

magnitudes of the original parameters.38 For all intents and

purposes, the CA1:1 value merely suggests little or no covalent

interaction between the (Et20)Cr2(tfa)4 adduct and donors.

The real quantities of interest for comparison to other dimer

systems are the acid parameters associated with the naked dimer,

Cr2(tfa)4 and are extricable from an analysis of the 2:1

enthalpies. The EAl:1 and CA1:1 values are defined by Eqns 2-2

and 2-3.

EA1: = EA k EB (2-2)

CA1:1 = CA k' CB (2-3)

Thus, determination of the inductive transfer parameters, k and

k', and using the base parameters for Et20 allows calculation of

EA and CA for Cr2(tfa)4. By substituting Eqns 2-2 and 2-3 into

the E and C equation, 2-1, the first and second enthalpies are

related to each other by32

AH2:1 = -AH1:1 -kEB2 k'CB2 (2-16)

In the present study, where an exchange reaction is being

studied, Eqn 2-16 takes on a slightly different form. Defining

all the enthalpy components AH1, AH3, AH5 and AH6 in terms of

their E and C components results in many cross terms which define

the perturbation that coordinated ether makes on the first

exchange reaction and that coordinated base, B, makes on the

second. The derivation is given in Appendix III and the

simplified solution has the familiar form

- AH2:1 = -AH1:1 k(EB EEt20)2 k'(CB CEt20)2 (2-17)

This equation has the same form as Eqn 2-1 and the ECW

program is used to solve for k and k'. The input data are the

heat differences as AAH, the squares of the quantities in

parentheses as the base parameters and W is, of course, zero.

The experimental heat differences, AAH = AH2:1 AHi:1, used

to determine k and k', along with those calculated from the best

fit parameters are given in Table 2-6. The attendant conditional

standard deviations are given in parentheses. The value for HMPA

was rejected in light of the large standard deviation and k and

k' were determined from the remaining five data points. The

best fit values (standard deviation) are k = 1.54 (0.13) and k' =

0.0079(0.0043). These are the parameters associated with the Cr-

Cr bond and demonstrate the ability to transmit electrostatic (k)

and covalent (k') effects. Equations 2-2 and 2-3 may now be used

to calculate the EA and CA values for the naked chromium dimer,

Cr2(tfa)4. The derived values are given in Table 2-7 along with

those for the molybdenum and rhodium dimers for comparison.

Again, the large EA and very small (negative) CA demonstrate

the pronounced Lewis acidity of the chromium dimer and its

tendency to interact with Lewis bases in primarily an ionic

fashion. This finding is consistent with the lack of

experimental success towards generating a dinuclear chromium

carboxylate free of axial coordination.11 In the absence of

donor molecules, the chromium carboxylates exist as polymers

where the very ionic carboxylate oxygens of neighboring dimers

serve as donors.

The low k' value which corresponds to transmission of the

base covalent parameters indicates that base binding does not

serve to polarize the bonding density of the chromium-chromium

bond, consistent with the poor orbital overlap between the two

metal centers. The rather large k value which corresponds to

transmission of the base electrostatic parameter is rather

surprising. Electrostatic interaction of the base dipole on one

side with the second metal center and with the second base dipole

should vary as 1/r2. Of the three systems studied, the chromium

dimer exhibits the longest metal-metal bond (2.54 X for44

Cr2(tfa)4(Et20)2, 2.39 X for73 Rh2Acq(H20)2 and 2.09 X for73

Mo2(tfa)4) and should exhibit the smallest value for k in the

series. For comparison purposes, it is probably more appropriate

to consider the relative values of k for each metal system. The

k values quantify the change in EA at the second metal center as

a consequence of bonding a donor at the first, irrespective of

Table 2-7. Acid parameters for various dinuclear carboxylates.

EA CA k k'
Rh2(but)4 5.06 1.74 1.16 0.0364

Mo2(pfb)4 5.92 0.385 1.46 0.022
Cr2(tfa)4 15.1 -1.54 1.54 0.0079

the magnitude of EA. The roughly twice as large EA for the

chromium system would display relatively half as large

perturbations on the magnitude of EA compared to the other two

dimers. This insight, then, is consistent with the longer metal-

metal bonds in the chromium dimers. Thus, more appropriate

measures of the ability of a metal-metal bond to transmit

coordination effects would come from k/EA and k'/CA instead of

comparing the absolute k and k' values.

d. Magnetic susceptibility

Theoretical investigations on dichromium tetraformate at the

SCF-level have predicted both the quadruple bond, 02,462, and

no bond, 02626*2 o2, configurations, neither of which

correspond to realistic descriptions.74 Incorporation of

correlation effects which allow mixing in of excited states in

the ground state description give more satisfying results28 in

which the 02 Tr462 term is important, but the ground state bond

order is closer to 1.5. Dichromium tetraacetate is diamagnetic75

and deMello et al. have proposed42 two antiferromagnetically

coupled chromium centers and no net covalency between the

chromium atoms as the dominant description for Cr24+ centers.

This latter description, too, suffers from inconsistencies by

predicting unrealistically long Cr-Cr distances. At shorter

(1.8-2.5 1) experimental bond lengths, d orbital overlap is

expected, and Zerner qualifies his calculations to suggest that

some degree of covalency accompany the antiferromagnetic


The trifluoroacetate bridged dimer studied here may represent

just such a borderline example with a chromium-chromium

separation of 2.54 A. An early magnetic study7 reported

magnetic moments of i eff= 0.74 BM for "Cr(F3CCOO)2" and 0.85 BM

for "Cr(F3CCOO)2.Ather". These species are presumably dimeric,

and the increase in magnetic moment upon ether complexation is

consistent with both an antiferromagnetic description and the

proposal that ligand donors weaken the metal-metal bond through

partial population of the o orbital.4677 The only other first

row transition metal for which a range of metal carboxylate

dimers has been prepared is copper. The cupric carboxylate

dimers exhibit long (2.6-2.9 A) metal-metal distances and

incomplete spin pairing between the two d9 centers. An

impressive array of these complexes78 has been prepared and

characterized magnetically and structurally, with a general goal

of determining the factors (structural parameters; nature of

bridge; axial ligands) and mechanism which contribute to spin

exchange. The following conclusions appear to be general for the

cupric carboxylate dimers:

1) the unpaired electrons reside in the dx2_y2 orbitals;

overlap is minimal at these metal-metal distances, ruling out any

direct exchange or covalency;

2) overlap with carboxylate Tr orbitals allows a super-

exchange pathway via the carboxylate bridges, resulting in

antiferromagnetic interactions with -2J ranging from 217 to

555 cm-l;

3) the singlet-triplet separation (-2J) is relatively

insensitive to the Cu-Cu distance;

4) -2J is sensitive to the Cu-0-C-O-Cu bridge distance and

angle, generally decreasing at longer distances;

5) -2J is sensitive to axial ligation, generally increasing

as the terminal ligands become stronger electron pair donors.

The chromium dimers, on the other hand, exhibit some degree

of orbital overlap, suggested to be incomplete in the

Cr2(tfa)4(Et20)2 adduct. The strong dependence of the Cr-Cr

distance upon axial ligation suggested magnetic studies to

complement the copper studies in treating a metal-metal bond with

some degree of covalency.

The room temperature magnetic susceptibilities of various

adducts of the form Cr2(tfa)4L2 were investigated by the solution

Evans method79 by generating the complexes in situ. This method

allows calculation of experimental magnetic susceptibilities by

measuring the proton chemical shift of an inert substance in the

presence of a paramagnetic complex. A coaxial tube arrangement

was used which contained a solvent system of 2% v/v C6H6 and 2%

v/v TMS in C6D6 in both the outer 5mm tube and the inner 1mm

capillary. The inner capillary also contained the paramagnetic

species at a concentration of ca. 5x10-2M. For an inert

substance (in this case, C6H6 and TMS) the shifts caused by the

paramagnetic substance when one employs a nonsuperconducting NMR

instrument are given by

AH = 2TT (2-18)
f AK
H 3

where AK is the change in volume susceptibility. With the

cocentric tube arrangement, two resonance lines will be obtained

for each standard, with the line from the more paramagnetic

solution lying at higher field. The mass susceptibility, X, of

the dissolved substance is given by Evans as

Y __v + Xe Xo(do-d )
X 3Av Xo + x(do-s (2-19)

where Av is the frequency shift separation in Hz, v is the

spectrometer frequency (99.55 MHz), m is the mass of substance

contained in 1 ml solution, Xo is the mass susceptibility of

the solvent, do is the density of the solvent and ds is the

density of the solution. Brault and Rougee have presented80 a

modified form of Eqn 2-19 to calculate the molar magnetic


X3Av 1000 + XM X (2-20)
XM 27Tv c D

where C is the molar concentration of the paramagnetic complex, M

is the molecular weight of the complex and XD is the

diamagnetic susceptibility of the paramagnetic complex. As

Desmond points out,81 the original Evans equation required noXD

term since complex dismagnetic susceptibility was compensated for

by including an equal concentration of ligand in the reference

solution. The Brault and Rougee equation is more general but

fails to account for density difference contributions. The

density term can often be neglected but becomes important when

measuring the small shifts for weakly paramagnetic substances.

Desmond includes the density term to obtain the equation

X 3Av 1000 + XoM D + xo(do-ds) 1000 (2-21)
XM 27Tr C C

The XD term is usually approximated from Pascal's constants82

and may be quite erroneous for large molecules. To alleviate

this hazard and the need for a density determination, the

diamagnetic correction may be obtained by NMR. Setting XM = 0

and moving the first two right-hand terms of Eqn 2-21 to the

other side gives an expression for the combined diamagnetic

susceptibility and density terms. An Evans experiment performed

on the diamagnetic analogue Rh2(tfa)4(THF)2 gave no peak shift

nor was any asymmetry evident in the TMS or C6H6 peaks. The sum

of the diamagnetic term and density correction then is -5.63 x

10- (obtained by setting both XM and Avequal to zero). The

calculated diamagnetic term is -3.0 x 104 (from Pascals

constants, where each Rh(II) is given a value of -20 x 10-6 ml

mol-"), and the difference approximates the density correction.

An upper error limit on the shift determination is about 0.3 Hz

which corresponds to a 0.43 x 10" contribution to the

diamagnetic/density term. Substitution of the value for the

combined complex diamagnetism plus density correction into Eqn 2-

21 gives

5 Av 1000__ -4
3 Av 1000 + XoM + 5.63 x 104 (2-22)
XM = 2 v c

At the concentrations used, complex precipitation warranted shift

determination in CD2C12 in several instances. Equation 2-22 was

used to calculate molar magnetic susceptibilities from both the

C6D6 and CD2C12 solution data. The single greatest source of

error in these measurements probably comes from determining the

shift magnitude. If the error calculated for the diamagnetic

correction is doubled to account for dual shift measurements (one

for the rhodium complex and one for the paramagnetic complex) the
attendant error is estimated to be 0.86 or about 1 x 10 .

Use of the "spin only" formula for the molar susceptibility

allows determination of the effective magnetic moment

eff = 2.84(X mT)1/2 (2-23)

where the product is expressed in Bohr magnetons. X M is a molar

quantity, calculated per mole of dimer, while Peff is per metal

and must be determined from 1/2 XM.

The molar magnetic susceptibilities and moments calculated

from the observed frequency shifts are reported in Table 2-8.

Benzene solution was the method of choice (economy), but some of

the dimer adducts were too unstable towards oxidation in this

solvent, necessitating determination in methylene chloride.

Immediately obvious is a very satisfying trend towards larger

susceptibilities and moments with increasing donor strength.

This relationship provides the first conclusive evidence for

genuine paramagnetism in a series of chromium carboxylates since

this consistency would not be observed with random trace

contamination by Cr(III) impurities.11

The measured susceptibilities roughly parallel the

experimental enthalpies except in one instance. The DMCA adduct

displays a larger moment than might be expected from the heat

Table 2-8. Magnetic susceptibilities and moments for Cr2(tfa)4L2










































































per dimer, + 1 X 10-4.

per Cr(II) atom, + 0.35 B.M.

multiple determinations.

determined in CD2Cl2.

data. Subsequent redetermination gave essentially the same

value, however, and both are reported.

Besides the chromium and widely studied copper systems

reports of two other discrete first row transition metal

carboxylate dimers have appeared, both of which display weak

antiferromagnetic exchange interactions. Dicobalt (II)

tetrabenzoate bis quinuclidine,83 Co2(C6H5CO2)4(quin)2, shows a

long Co-Co distance of 2.83 1 and the high spin Co(II) centers

display weak antiferromagnetism of -2J = 38 cm-1. At this long

metal-metal distance, spin coupling is expected to occur via the

bridge super exchange pathway. A vanadium complex,23

V2(tfa)4(C5H5)2, exhibits a V-V distance of 3.7 A and weak

antiferromagnetism, suggested also to proceed through a super

exchange path involving the carboxylate bridges.

Treatment of the exchange mechanism operative in the chromium

dimers is necessarily more complex than for the copper

carboxylates,8 the former involving S = 2 metal centers. MO

calculations support direct orbital overlap and theoretical

treatment would include both direct and super exchange pathways.

Calculations showing low lying excited states do not rule out the

possibility of thermal equilibrium between the populations of

ground level and the first excited level with other multiplicity

(singlet-triplet equilibrium).85 The latter has not been

demonstrated in the binuclear carboxylates but occurs in

monomeric complexes when the ligand field splitting is close in

energy to the electron pairing energy. The trend towards larger

moments with increased ligand field, however, is not consistent

with a singlet ground state.

Determination of the singlet-triplet splitting (-2J) can only

come from measurement of the temperature dependence of the

magnetic susceptibility, and these data are not yet available for

the chromium trifluoroacetate dimers. The data in Table 2-8 do,

however, suggest a direct exchange mechanism. In modeling the

magnetic susceptibility data for some 140 compounds of the

binuclear copper carboxylates, Jotham and Kettle observed a

general trend for J to increase as either the terminal or the

bridging ligands become better electron donors.86 At a given

temperature, larger values of -2J would give smaller experimental

moments. Thus in these systems in which the super exchange

pathway predominates, better donors give smaller magnetic

moments. The reverse is true for the chromium series studied

here. Donor lone pairs interact primarily with the chromium

dimer o orbitals. Stronger donors would serve to destabilize

the Cr-Cra orbital while stabilizing the Cr-Cr o* orbital, and

partial population would weaken the covalent bonding, consistent

with the observed trend for larger moments with better donors.

e. EPR, Electrochemistry

The cyclic voltametry of Cr2(tfa)4(Et20)2 was attempted with

the aim of finding the proper conditions for controlled oxidative

electrolysis. An EPR investigation of the cationic product might

add insight into the nature of the chromium-chromium bond. The

EPR prarmeters would be descriptive of the HOMO orbital which

surrendered the electron. In methylene chloride, the dimer was

found to be incompatible with the electrolytes Bu4NBF4, Bu4NPF6

and Bu4NI, all of which reacted to give blue or green solutions.

No reaction was observed with BuqNCIO4. No oxidation wave could

be detected in solution, however, with this electrolyte though a

slow irreversible reduction occurred at -0.45 V (Ag/AgCl). No

further attempts were made to oxidize the complex.

The magnetic studies suggested unpaired electron density at

the Cr(II) centers, prompting investigation by X-band EPR. The

solid state adduct Cr2(tfa)4(Et20)2 displayed no EPR down to 10K.

In CH2C12 solution, the ether and HMPA adducts gave no signal

down to 85K. One attempt was made to record the solution

spectrum of Cr2(tfa)4(Et20)2 at liquid helium temperatures. A

1x10-2M CH2C12 solution glass gave the spectrum in Fig 2-9. The

observed asymmetric signal displays no 53Cr (9.5%) hyperfine

splitting and would appear to be that for a system of axial

symmetry with gL = 1.98 and g11 = 1.87. A forbidden S = 2

transition may be present at H/2 (1600 G) but this is speculative

in lieu of experimental clarification. Lack of any observed

signal for a non-dilute powder sample at the same temperature is

consistent with significant spin lattice relaxation while lack of

any EPR at 77K (methylene chloride glass) would indicate

extensive spin-orbit coupling. EPR in d4 systems is very rare

due to short spin lattice relaxation times and a large zero

field splitting. A comprehensive review of the literature,87

albeit in 1972, yielded only five d4 systems for which an EPR had

been reported; three of which were Cr(II) as dilute ionic salts.

The estimated g values reported here do not, however, resemble

any of those reported for Cr(II) or Cr(III) systems. This




L .



w c


"-I m

p, 0

finding warrants further investigation and is reported here only

as a matter of record.

3. Further Reactivity

Chromium (II) salts have been extensively employed as

reducing agents in preparative organic chemistry.88 Chromium

(II) chloride, sulphate and perchlorate are similar in their

scope; chromium (II) ethylenendiamine cation functions as a more

efficient reducing agent while chromous acetate is a milder

reducing agent, reacting under relatively neutral conditions. By

contrast, the utility of chromous trifluoroacetate as a reducing

agent remains relatively unexplored and the ligand exchange

reactions explored here offer some insight into its redox


Reduction of alkylhalides by chromium (II) salts to generate

olefins, carbenes, or alkylchromium species via haloatom

abstraction represents one of the best known applications of

these reagents. Methylene cloride solutions of chromous

trifluoroacete were found to be stable for several months under

dinitrogen as evidenced by retention of the robust purple color.

Irradiation (550 nm, 6-+- T' ) of solutions for 30 minutes in the

UV-VIS beam path also showed no appreciable spectral changes.

As mentioned for the case with DMA, UV-VIS titrations were

well behaved, defining a two step exchange process. Deviation

from the 2:1 limiting spectrum with excess base (> 5 equiv.) was

apparent in all cases and was accompanied by an immediate blue to

green color change which became more pronounced with additional

added base. FTIR titrations revealed a new vco2,asy at 1705

to 1710 cm-1 with excess (> 2 equiv.) donor which was assigned to

monodentate trifluoroacetate. Chromium (II) oxidation would thus

appear to be facilitated by bridge removal and equatorial donor

coordination. Additionally, except for Cr2(tfa)4L2 species

where L = Et20, DMTF or Et3PO4, the final blue solutions from

both spectral and calorimetric titrations turned green within 2

to 24 hours despite rigorous exclusion of dioxygen, indicating

oxidation when excess donor was present in solution. These

observations are consistent with electrochemical measurements on

a series of tetracarboxylato-dirhodium (II) complexes.89 Das et

al. found that electron withdrawing substituents on the

carboxylate bridges produced less easily oxidized dimers. In

their studies, the lower oxidation state of Rh2(tfa)4 was

stabilized to such an extent as to see no oxidation step within

the solvent limits. A similar trend in solvent effect was

observed with coordinating solvents such as pyridine and DMSO

giving the most negative oxidation potentials. The chromium

dimers are stable indefinitely in solution if stoichiometric

amounts of base are present, allowing ready isolation of the

exchange species.

A crystal structure has been reported53 for the reaction

product of the strong donors pyridine and 4-cyanopyridine with

Cr2(02CCF2H)4(Et20)2. The isolated crystals are trichromium (II,

III, III) compounds with the basic iron acetate trimer structure.

No spectral results were reported. The same procedure was

followed using the trifluoroacetate and pyridine to give long (up

to 2 cm) pale green needles which turn dark olive upon exposure

to air. Assuming the same structure, the bridging

trifluoroacetates show a Vco2,asy at 1671 cm- in the mixed

valence trimer. Attempts to induce trimer formation with the

donors DMA and DMCA were unsuccessful, surrendering no solid

products after three weeks. Both the blue DMA-containing and

dark green DMCA-containing solutions gave Vco2,asy peaks at

1717 and 1713 cm respectively, indicating monodentate

trifluoroacetate. Trimer formation does not appear to be

general, with most donors simply inducing slow oxidation of

chromous trifluoroacetate, and the effect is more pronounced in

benzene than in methylene chloride.

C. Conclusion

A thorough investigation of the coordination chemistry of

chromous trifluoroacetate has been performed with the aim of

understanding the transmission of bonding effects through the

weak Cr-Cr bond and the perturbation of donor ligands on the

metal-metal bond. Spectral titrations indicate that donor

ligands decrease the d orbital overlap between the metal centers.

Calorimetric studies show a lower enthalpy for second base

coordination resulting from transmission of donor effects from

one metal center to the next. The effect is less pronounced than

in the previously studied Rh2(but)4 and Mo2(pfb)4 systems. A

correlation analysis of the calorimetric data allows description

of the Cr(II) Lewis acidity as a relatively strong acid,

interacting in primarily an electrostatic fashion with Lewis

bases. Excellent data fits support an inductive transfer model

used to describe communication between the metal centers.

Magnetic susceptibility measurements on a range of Cr2(tfa)4L2

adduct species reveal a pronounced influence by donors upon

dimer paramagnetism and support a direct exchange pathway for

spin pairing. UV-VIS, NMR and IR data support the donor exchange

reactions proposed and reveal a destabilization towards

oxidation when stronger donors coordinate to the Cr2 center.

D. Experimental

1. Data Analysis

Programs. The following computer programs were used for

analysis of raw spectral and calorimetric data.


Written by J. R. Long,68 this program utilizes a standard non

linear least squares routine to provide the best A6(EI 0) and

K which describe a 1:1 equilibrium, A + B 4AB. Raw data needed

are concentrations and absorbance changes for a spectral

titration. Raw heats can be used to solve for AH and K for a

calorimetric titration.


Written by T. Kuechler,68 SPEC utilizes concentrations and

spectral changes to solve for the best K1 and K2 via least

squares for a two step equilibrium. The large number of unknowns

solved for (K1, K2, and E0, El' 2, at each wavelength used)

requires good initial estimates of each of these parameters to

avoid false minima. Useful algorithms for initial estimates are

provided by J. R. Long.68 In practice, best results are obtained

by fixing most of the parameters (usually the e's) and allowing

the computer to vary those that are least well known (usually the



Written by J. R. Long,68 this least squares program solves for

the best fit molar enthalpies, AHI and AH2, for a two step

equilibrium. Input includes concentrations, raw heats and

equilibrium constants for a calorimetric titration.

Error analysis. Output parameters from the above programs

are provided with marginal and conditional standard deviations,

MSD and CSD, which demonstrate how well the model fits the data.

The conditional standard deviation defines the magnitude of error

while the MSD/CSD ratio indicates how well defined the parameter

is.68,90 For a reasonable CSD, results are considered meaningful

if the ratio is less than 4, tentative if the ratio is between 4

and 12, and not meaningful if the ratio is larger than 12.

ECW Program. Revised by M. K. Kroeger, this program uses

experimentally determined enthalpies to calculate the best fit

values of the unknown parameters EA, CA (and W if necessary) to

the following equation.


The EB and CB are found in references 38 and 39. Alternatively,

measured heats and and known EA, CA pairs can be used to solve

for EB, CB. A description of the computer program has been

previously reported.91

2. Materials

Metal complexes. All syntheses and manipulations were

performed under dinitrogen using Schlenk techniques or an inert

atmosphere box. Chromous complexes are generally quite oxygen

sensitive and the compounds used here are oxidized within seconds

upon air exposure.

Dichromium (II) tetrakistrifluoroacetate bisdiethylether, I.

Chromous acetate was generated from Zn/Hg reduction of Cr(III)

(aq) and sodium acetate.92 Yellow chromous carbonate was

synthesized through exchange of the acetate bridges by reaction

with potassium carbonate in water.93 Chromous trifluoroacetate

was prepared by a modified literature procedure.9 Under

dinitrogen, 5.5 g (9.9 mmol) K4Cr2(CO3)4(H20)3 and 8.0 ml (100

mmol) trifluoroacetic acid were refluxed in 80 ml deaerated

diethylether for six hours. The Schlenk flask was plunged

briefly into a dry ice bath, the purple ether layer decanted from

the frozen blue aqueous layer and the ether removed by vacuum.

Extraction with benzene followed by two recrystallizations from

benzene and vacuum drying (30 min, 250) produced purple blocks of

Cr2(tfa)4(Et20)2. Prolonged evacuation (>6 hr) was found to

strip off coordinated ether. Final yields were typically about

20%. Attempts to purify by sublimation decomposed the complex.

Repeated elemental analyses typically showed loss of 5-10%

coordinated diethylether. Calculated for C16H20Cr2F12010:

C, 27.29; H, 2.86; Cr, 14.77. Found: C, 25.91; H, 2.82;

Cr, 15.01.

Dichromium(II) tetrakistrifluoroacetate bisdimethylthioform-

amide. I, 34 mg (0.048 mmol) was dissolved in 1 ml benzene.

Addition of 9.0 1p (0.11 mmol) DMTF produced a clear solution and

a fine blue powder within ca. 5 min. The solid was filtered,

washed once with cyclohexane and vacuum dried (250) one hour.

Dichromium(II) tetrakistrifluoroacetae bis dimethylformamide.

I, 37 mg (.053 mmol) was dissolved in 1 ml benzene. Addition of

9.0 p1 (0.12 mmol) DMF produced a clear solution and a fine blue

powder within ca. 5 min. Solid was filtered, washed once with

cyclohexane and vacuum dried (250) one hour.

Dichromium(II) tetrakistrifluoroacetate bis trimethylphosphite

I, 33 mg (0.047 mmol), was dissolved in 1 ml benzene. (MeO)3P, 12

Ip (0.10 mmol), was added to give a slightly bluer solution.

Stripping off the solvents by vacuum produced a blue violet

powder which was not treated further.

Dichromium(II) tetrakistrifluoroacetate bis tetramethylurea.

I, 222 mg (0.32 mmol), was dissolved in 20 ml methylene chloride

and 76 pl (0.64 mmol) tetramethylurea added to give a

blue solution. Solvents were stripped off, the solid extracted

with benzene and the volume reduced to 5 ml. The solution was

suspended in a dewar above ice. The rectangular dark blue

crystals which formed after three days were collected and vacuum

dried (250) one hour. This compound slowly decomposes in the

solid state, turning green after about one month under


Dichromium(II) tetrakistrifluoroaectate bis Triethylphosphate.

I, 490 mg (0.70 mmol), was dissolved in 20 ml benzene and 0.35

ml (2.1 mmol) triethyl phosphate added to give a deep blue

solution. Volume was reduced to 5 ml and cooled to 5 for 24

hours to produce a mass of dark blue cubic crystals. Solid was

collected, washed with cyclohexane and vacuum dried (250) 30 min.

Calculated for C20H20Cr2F12016P2: C, 26.10; H, 3.29; F, 24.77.

Found: C, 26.11; H, 3.40; F, 24.44.

Dichromium(II) tetrakistrifluoroacetate bis hexamethylphosphor-

amide. I, 525 mg (0.74 mmol), was dissolved in 20 ml benzene and

0.30 ml (1.7 mmol) hexamethylphosphoramide added to give a deep

blue solution. Solution was cooled to 5 and bright blue cubic

crystals began to form after 15 min. After 24 hours, crystals

were filtered and dried under a stream of dinitrogen. Calculated

for C20H36Cr2F12N6010P2: C, 26.27; H, 3.97; F, 24.93; N, 9.19.

Found: C, 26.34; H, 3.99; F, 24.38; N, 9.21.

Attempts to isolate solid adducts of dimethylacetamide and

dimethylsulfoxide by similar procedures in benzene were

unsuccessful, producing a pale green gel in both instances.

Cr3O(02CCF 3)(C H N) I, 140 mg (0.20 mmol), in 5 ml Et20 was

placed in a test tube. 4 ml of hexane was layered onto the ether

and 0.8 ml (9.9 mmol) pyridine in 8 ml hexane added to the hexane

layer. After four days, solutions were almost clear. Large (1 X

1 X 20mm) pale olive needles had grown down from the solvent

interface. A much smaller yield of small dark olive crystals

were sparsely formed on tube walls near the bottom. Crystals

were collected and dried (vac., 250) one hour. Only the pale

green complex changes color (dark olive) upon exposure to air.

Bases, solvents. Methylene chloride was dried over CaC12

and vacuum distil led from P205 at 250. Benzene was dried over

CaC12 and vacuum distilled from CaC12 at 250. Pyridine was dried

over Na and fractionally vacuum distilled from BaO. Diethyl

ether was vacuum distilled from Na/benzophenone at 250. DMSO was

dried over NaOH and fractionally vacuum distilled from NaOH. DMF

was dried over KOH and fractionally vacuum distilled from BaO.

DMA was dried over molecular sieves and fractionally vacuum

distilled from BaO. DMTF was dried over BaO and fractionally

vacuum distilled. DMCA was fractionally vacuum distilled. HMPA

was fractionally vacuum distilled from BaO. Trimethyl phosphite

was fractionally vacuum distilled. Trimethyl phosphine oxide was

used as received. Tetramethyl urea was dried over BaO and

fractionally vacuum distilled. Triethyl phosphate was

fractionally vacuum distilled from BaO. DMTF was dried over BaO,

fractionally vacuum distilled and stored in the dark. All

distillations were performed under dinitrogen or vacuum and only

the heart cut saved.

3. Data Collection

All manipulations were performed under dinitrogen in an inert

atmosphere box or employing syringe techniques. All glassware,

and cells were stored in dessicators over CaSO4 or in the dry


Electronic spectroscopy. UV-VIS titrations were performed by

repeated microliter injections of a concentrated stock base

solution into the standardized 5 ml chromium solution. The

quartz inert atmosphere cell has been described.95 Methylene

chloride solutions were prepared in volumetric flasks inside the

inert atmosphere box. Spectra were recorded on a Perkin Elmer

330 spectrophotometer.

Calorimetry. Design and operation of the calorimeter has

been previously described.96 To correct stability problems, a

new unit was constructed in the Electronic Shop of the University

of Florida chemistry department. Schematics and operating

procedure are given in Appendix II. Typically, a methylene

chloride solution (ca. 10-2M) of the dimer was prepared in a 50

ml volumetric flask and decanted into the inert atmosphere 55 ml

adiabatic cell. Five milliliters of solvent were syringed into

the volumetric flask and washed into the cell. A 5 ml aliquot of

solution was removed from the cell and transferred to the UV-VIS

cell. Concentrations were calculated from the absorbance at 550

nm (X550 = 132.5). The base solution (ca. 0.5 M) was prepared in

a volumetric cell and used to charge a 1 ml Hamilton gas-tight

syringe. After expelling any dead volume and air bubbles, the

syringe was emptied to the zero calibration stop and secured into

the cell. While stirring, the cell and contents were allowed to

equilibrate to room temperature (250) for several hours. A

series of thirteen calibrated syringe stops allowed incremental

injection of known volumes of base solution. Recorder

deflections which accompanied the exothermic base additions were

ratioed to a calibration deflection of known heat. Identical

titrations were performed in the absence of metal complex to

measure the base solution dilution heats (if any). Measurable

dilution heats were not observed with any of the donor solutions.

FTIR. Methylene chloride solutions were scanned in airtight

0.2 mm solution cells, after preparation in the glove box.

Solvent absorbances were subtracted. Spectra were recorded on a

Nicolet 5DX-B instrument.

FT-NMR. CD2C12 and C6D6 solutions were prepared in the inert

atmosphere box and 100 MHz spectra recorded on a JEOL XL-100 FT-

NMR. Determinations of magnetic susceptibilities with the EVANS

method were performed by modified procedures97 with a newly

designed coaxial tube arrangement, shown in Fig 2-10. Chromium

solutions of 1-5 X 102M were prepared by addition of 2.2-3

equivalents of the donor of interest to C6D6 solutions of


1 d

teflon spacer

paramagnetic solution
in Imm i.d. capillary


Fig 2-10. Modified NMR tube for determination of solution
magnetic susceptibilities of air-sensitive complexes.

Cr2(tfa)4(Et20)2. Based on the large calculated equilibrium

constants, the added donor effectively displaces coordinated

ether. The same solvent system, C6D6 containing 2% v/v TMS and

2% v/v C6H6 was used for both tubes. Paramagnetic solutions were

syringed into clean dry melting point capillary tubes (1 mm i.d.)

and sealed by fitting with a machined teflon spacer. A tight

fitting cap provides a second oxygen barrier and the NMR Spectra

were collected within 15 minutes of sample preparation. The dual

standards provide a check on locating the paramagnetic shifted

resonance and signal integration provides another. This tube

arrangement allows facile manipulation of the air sensitive

solutions and avoids the expense of coaxial tubes.

Elemental analyses. Analyses were performed by Galbraith

Laboratories, Knoxville, TN. The expense of routine analysis of

the air sensitive complexes prohibited characterization of each

adduct species. The HMPA and Et3PO4 ( a strong and intermediate)

donor complexes were characterized as representative complexes.



A. Introduction

Trans influence--the ground state influence on a metal-ligand

bond strength by another trans coordinated ligand--plays a

fundamental role in coordination chemistry and transition metal

complex reactivity. The ability of coordinated ligands to

labilize or stabilize a trans coordination site serves to define

the metal coordination sphere. A general feature of trans ligand

influence in mononuclear chemistry occurs when both metal-ligand

bonds are in competition for the same metal d orbital.98

The idea of a trans influence in dinuclear complexes has been

vigorously pursued, and, in a general sense, the metal-ligand

bond strength has an inverse relationship with the metal-metal

bond strength. As the studies in the last chapter illustrate,

the effect is pronounced in the chromium carboxylates as both

ligand and adjacent metal atom compete for the same chromium

orbital; stronger chromium-ligand bonds result in weaker metal-

metal bonds. Conversely, the strong quadruple Mo-Mo bonds in

the molybdenum carboxylates dictate weak metal-ligand

interactions. The effect is schematically illustrated in Fig 3-1.

While structural studies of dimers with stronger metal


L >M =>M<- L

Fig 3-1. Schematic representation of the trans influence in
metal carboxylate dimers.

LI ,

M t-- L

interactions illustrate slight lengthening of the metal-metal

bond with ligand donor strength, the overall effect is a

weakening of the second metal-ligand interaction as demonstrated

by calorimetric measurements.32 Redistribution of electron

density donated at one metal over both metal atoms lowers the

Lewis acidity of the second, and the effect has been

parameterized in terms of communication between the two metal

centers. As a further test of this model, the studies reported

here examine the effect of varying one ligand on the spectral

properties of a second, fixed, trans ligand in a rhodium dimer.

Examples of mixed ligand and corresponding fixed ligand complexes

for comparison are scarce. The results reported here provide the

first systematic investigation of the trans influence across a

metal-metal bond. First the EPR parameters of a coordinated spin

label (TEMPO) in complexes of the type (L)Rh2(pfb)4(TEMPO) are

examined, followed by the stretching frequency of coordinated CO

in complexes of the type (L)Rh2(pfb)4(CO). The original

experimental results have been communicated49 by James Stahlbush;

a reinvestigation of the data is offered here.

Relationships between spectral and bonding properties of 1:1

adducts have been reported,99-101 such as the linear correlation

between AvOH and AH for a series of 1:1 phenol-base adducts.

Breakdown of this correlation upon extension to a larger donor

set is attributed to fundamental differences in AvOH and A H for

gauging E and C effects. A more appropriate treatment of

spectral changes which accompany adduct formation involves an

adaptation of Eqn 2-1. Replacing AH by AX, the spectral shift,

gives101 Eqn 3-1 for the case where a base is held constant and a

series of acids studied.

AX +W = EAEB + CACB (3-1)

The asterisks imply that conversion units for converting EA from

(kcal mol")1/2 are included in EB along with the response to

the quantity being measured induced in the base by the acid.

The analysis of spectral data via Eqn 3-1 serves several

purposes. Besides establishing a correlation for a particular

acid or base which allows prediction of bond strengths from

spectral shifts, this treatment extends the E and C basis set

beyond only enthalpy data. And as with the enthalpy data,

correlation of the easier to obtain spectral shifts lends insight

into the nature of the metal-metal and metal-ligand bonds.

B. Results and Discussion


a. EPR spectra

Species of the type (B)Rh2(pfb)4(TEMPO), where B is a

coordinated Lewis base and TEMPO is 2,2,6,6-tetramethyl-

piperidine-N-oxyl, were investigated using EPR in order to

measure the influence of B on the EPR spectrum of the coordinated

nitroxyl radical. TEMPO is a donor of moderate strength that

does not bind to Rh2(OAc)4 or Rh2(but)4 but forms adducts with

the corresponding fluorinated derivatives, Rh2(tfa)4 and

Rh2(pfb)4. Typically, solutions were prepared in a 9:10:1 molar

ratio of base: rhodium complex: TEMPO in oxygen-free methylene

chloride. Most of the nitroxide existed in the unbound state,

while most of that coordinated existed as the (B)Rh2(pfb)4(TEMPO)

species. A representative spectrum for dimethylacetamide with

Rh2(pfb)4 and TEMPO is shown in Fig 3-2. In all cases, the

signal for the (B)Rh2(pfb)4(TEMPO) species appears between those

for the free nitroxide and Rh2(pfb)4(TEMPO). The nitrogen

hyperfine for these 2:1 adducts was usually either equal to or

slightly less than that observed for the Rh2(pfb)4(TEMPO) adduct.

Both of the above effects would be expected from an inductive

weakening of the rhodium nitroxide bond by the coordinated base.

That is, a coordinated donor weakens the Rh-TEMPO bond, and the

EPR parameters of the spin label move towards those of free

TEMPO. A wide range of g-values was observed for the mixed donor

2:1 adducts while only minor or no changes were observed for the

nitrogen hyperfine, AN. A tabulation of the EPR parameters is

given in Table 3-1.

Bonding in Rh2(tfa)4(TEMPO) has been described in terms of o-

donation from a nitroxide oxygen lone pair into the Rh-Rh a

orbital with concomitant orbital mixing of the Rh-Rh 7T orbital

with the nitroxide T* containing the unpaired spin. When B is a

sigma donor, the metal nitroxide bond is weakened, and the g

value moves toward that of free nitroxide. When B is a pi

acceptor the metal nitroxide bond can be weakened in two ways.

Directly, competition for metal-metal 7T electron density

decreases the Tr-backbonding to TEMPO. Indirectly, a pi acceptor

exhibits enhanced sigma donation, causing weakening of the metal

nitroxide bond in the same way as a pure sigma donor. The net

result is that both o-donors and T-acceptors serve to weaken the

metal nitroxide bond and cause a lowering of the g-value back

towards the free solution value of TEMPO.


l .... 4 -i i 1111 IIIII.IIIIH n


20 G


Fig 3-2. Representative spectrum observed for a CH2C12 solution
of Rh2(pfb)4, TEMPO and dimethylacetamide. Species are a) free
TEMPO, b) (DMA) Rh2(pfb)4(TEMPO), c) Rh2(pfb)4(TEMPO), d)
precipitated Rh2(pfb)4(TEMPO).

Table 3-1. EPR Parameters of (B)Rh2(pfb)4(TEMPO) adducts.a

-AH 1:1

no base

methyl acetate

ethyl acetate



dimethyl acetamide

bridged there




dimethyl formamide



cage phosphitef






































































(kcal mol"1)





















Table 3-1 (continued)

a) Compare to solution parameters of free TEMPO: g = 2.0047, AN
= 1.47 X 10-3cm1.

b) g of (B)Rh2(pfb)4(TEMPO).

c) g calc from equation 3-4 using the C 1:1 and EA 1:1 from
equations 3-2 and 3-3 and values reported in Table 3-2.

d) Calculated enthalpy for adding B to Rh2(pfb)4 using values in
Table 3-2.

e) 7-Oxabicyclo [2.2.11 heptane.

f) 1-Phospha-4-ethyl-2,5,7-trioxabicylo [2.2.11 octane.

g) Systems in which metal T* to ligand T ~ back-bonding

h) Unresolved.

b. Quantitative correlations

The general trends discussed in the previous section suggest

that g-values from the EPR spectra may be used to provide

quantitative data about the strength of binding. This encouraged

us to investigate the quantitative relationship between the

enthalpys of adduct formation and changes in the g-values. A

model has been proposed and tested for predicting the enthalpy of

coordination of a second donor, B, to an M2(02CR)4(B) adduct to

form a 2:1 adduct.31-33 In this model, the EA parameter for the

1:1 adduct behaving as an acid to form a 2:1 adduct, EA1:1, is

given by

EA1:1 EA kEB (3-2)

and CA1:1 is given by

CA1:1 = CA -k'CB (3-3)

Where k and k' reflect (for a-donors) the effectiveness of the

metal-metal bond at transmitting the inductive influence of base

coordination to the second metal center. EA1:1 and CA1:1 values

for(B)Rh2(pfb)4 can be calculated from equations 3-2 and 3-3.

The reported32,38,39 E and C values used in this analysis are

given in Table 3-2. The EA11 and CA1:1 of the various

(B)Rh2(pfb)4 adducts are calculated and also listed in Table 3-2.

Note how the acid parameters decrease as the inductive effect

increases. That is, the greater the donor strength, the lower is

Table 3-2. E and C Parameters for Species Used in This Study

B EB CB EA 1: CA1:

methyl acetate 0.903 1.61 4.01 1.68

ethyl acetate 0.975 1.74 3.93 1.68

acetone 0.987 2.33 3.92 1.66

p-dioxane 1.09 2.38 3.80 1.65

dimethyl acetamide 1.32 2.58 3.53 1.65

bridged ether 0.887 4.11 4.03 1.59

tetrahydrofuran 0.978 4.27 3.93 1.58

dimethylsulfoxide 1.34 2.85 3.51 1.64

hexamethylphosphoramide 1.52 3.55 3.30 1.61

dimethyl formamide 1.23 2.48 3.63 1.65

acetonitrile 0.886 1.34 4.03 1.69

pyridine N-oxide 1.34 4.52 3.51 1.58

cage phosphite 0.548 6.41 4.42 1.51

diethylsulfide 0.339 7.40 4.67 1.47

4-picoline 1.17 6.80 3.70 1.49

pyridine 1.17 6.40 3.70 1.51

1-methylimidazole 0.934 8.96 3.98 1.41

piperidine 1.01 9.29 3.89 1.40

triethylamine 0.991 11.09 3.91 1.34

TEMPO 0.915 6.21 4.00 1.51

A EA CA k k'

Rh2(pfb)4 5.06 1.74 1.16 0.0364

the Lewis acidity of the second metal center. With these E and C

values for the various 1:1 adducts, we are now in a position to

attempt a correlation of the g-values obtained when TEMPO is

coordinated to the second coordination site to form a series of

2:1 adducts of general formula (B)Rh2(pfb)4(TEMPO). Substitution

of Eqns 3-2 and 3-3 into 3-1 gives

g + W = EA EB + CA :1CB (3-4)

where g has been substituted for AX The simultaneous equations

are solved for EB* and CB* which are the spectroscopic parameters

for TEMPO needed to predict g. The quantity W includes the g

value for free TEMPO (2.0047) as well as any nonzero enthalpy

components of the spectroscopic relation.102-103 The best fit

results yield

EB = 1.16 x 10-3 (0.29 x 10-3)

CB = 1.78 x 10-2 (0.10 x 10-2)

W = -1.9784 (.0018)

with standard deviations in parentheses. These parameters for

TEMPO allow calculation of the g-value for any

(B)Rh2(pfb)4(TEMPO) complex when the base is a o-donor whose EB

and CB parameters are known. Data for adducts with donors known

to act as Tr-acceptors acetonitrilee, cage phosphite,

diethysulfide, 4-picoline and pyridine) are not included in the

calculation of the TEMPO parameters. Attempted fits which

include the Tr-acceptors give larger standard deviations, as

expected. Table 3-1 contains the g-values calculated from this

fit (gcalc). The columns of g and gcalc show excellent

agreement, generally within the accuracy of the measured numbers,

except in those cases where metal to base Tr-backbonding occurs

(data in parentheses). For the latter, these are the g-values

expected if the Lewis bases, B, utilize only their a-bonding

capabilities in forming the (B)Rh2(pfb)4 adducts. The close

agreement between g and gcalc demonstrates that the inductive

model (equations 3-2 and 3-3) adequately describes the

transmission of coordination effects through the metal-metal

bond, for it is this model which describes the varying acidity of

the second metal center. For the five donors which also behave

as T-acids, the lower observed g-values than calculated by Eqn 3-

4 manifest the metal to B ir-backbonding contribution in these

adducts. Thus this analysis would suggest that both and 7T

interactions in the B-Rh bond serve to weaken the Rh-TEMPO bond.

To see if a relationship exists between g and the strength of

B binding to Rh2(pfb)4, enthalpies for binding bases were

calculated from the E and C equation (2-1); these 1:1 adduct

heats, AHB ,:1 are given in Table 3-1. The E and C parameters

are derived from a-only interactions and hence the calculated AH

in Table 3-1 reflect only the component of the adduct bond.

The experimental g-values, however, reflect the sum of o-donor

and any r-acceptor interactions. In Fig 3-3, the enthalpies of

1:1 adduct formation for (B)Rh2(pfb)4 are plotted as a function

of the experimental g values for the 2:1 adducts (B)Rh2(pfb)4(TEMPO).

Both calculated (o,A) and experimental (X) enthalpies are

included. (It would perhaps be more direct to compare the g-

values of the 2:1 adducts with the enthalpy of 2:1 adduct

formation, that is, the enthalpy for B + Rh2(pfb)4(TEMPO) or even

TEMPO + Rh2(pfb)4(B). The qualitative conclusions are, however,

the same, and experimental heats are available for B +

Rh2(pfb)4.) In Fig 3-3, the calculated (a-only) enthalpies are

all lower than the g-values would suggest for the five donors

which act as ir-acceptors: acetonitrile, cage phosphite,

diethylsulfide, 4-picoline and pyridine. Additional

stabilization in the (B)Rh2(pfb)4 adduct bond is consistent with

Rh to B fr-backbonding. In the two cases where experimental

AHB1:1 are available acetonitrilee and pyridine), the measured

heats lie much closer to the correlation line. Thus, the g-

values manifest both a and r effects across the metal-metal bond,

both serve to lower the g-value of coordinated TEMPO, and a-

donation appears to exert a stronger influence than 7-acceptance.

(Inclusion of T-effects brings AHBg:1 closer to the correlation

line but not all the way.)

2. Carbon Monoxide

a. FTIR spectra

The CO ligand is ubiquitous in organometallic chemistry, and

considerable effort has been put forth to understand the nature

of M-C-O bonding and the influence of various ligands upon the

reactivity and spectroscopic properties of carbonyls. In probing




o 15

10- 0


O o DONORS, -AHcalcd
x -H Hexpt

0 i I I I in
2.0070 2.0090 2.0110 2.0130 2.0150


Fig -. Correlation of predicted enthalpy of adduct formation,
AHB for Rh2(pfb)4 and donor, B, to the observed g value of
(B)Rh2(pfb)4(TEMPO). o,o-donors; A,T-acceptors. For those
systems where AHB11 has been experimentally determined, the
corresponding experimental heats are given by X. Character size
indicates range of experimental error. The best fit line is for
a -donors (see equation 3-4).