electronic and catalytic properties of iron porphyrin ...electronic and catalytic properties of iron...
TRANSCRIPT
Electronic and catalytic properties of iron porphyrin
complexes: Trends and reaction mechanisms.
A thesis submitted to the University of Manchester for the degree of
Doctor of Philosophy (PhD)
in the Faculty of Engineering and Physical Sciences
2015
Mala Alhaji Sainna*
Supervisor: Sam P. de Visser
School of Chemical Engineering and Analytical Science
2
Table of Contents LIST OF FIGURES ................................................................................................................................ 5
LIST OF SCHEMES ............................................................................................................................. 11
LIST OF TABLES ................................................................................................................................ 13
ACKNOWLEDGEMENT .................................................................................................................... 14
ABSTRACT .......................................................................................................................................... 15
DECLARATION .................................................................................................................................. 16
COPY RIGHT STATEMENT .............................................................................................................. 17
CHAPTER 1 ......................................................................................................................................... 18
Introduction ........................................................................................................................................... 19
1.1 Preface................................................................................................................................... 19
1.2 Homogeneous Catalysis ........................................................................................................ 21
1.2.1 Ligand effects. ............................................................................................................... 23
1.2.2 Bite angle effect ............................................................................................................. 24
1.2.3 Cone angle effect ........................................................................................................... 26
1.3 Ligand according to donor atoms. ......................................................................................... 27
1.3.1 Imido and Alkoxy groups as an anionic ligand. ........................................................... 27
1.3.2 Neutral and Anionic hydrocarbyl groups ..................................................................... 28
1.3.3 Oxazolines, imines , Amines and related ligands .......................................................... 28
1.3.4 Carbon monoxide and Carbene .................................................................................... 30
1.4 Common anions .................................................................................................................... 32
1.5 Vacant site Creation and coordination of the substrate ......................................................... 33
1.6 Insertion versus migration ..................................................................................................... 34
1.6.1 β-Elimination and de-insertion: .................................................................................... 39
1.6.2 Oxidative addition and Reductive elimination: ............................................................. 41
1.7 Activation of a substrate toward nucleophilic attack ...................................................... 41
1.7.1 Role of Alkenes and alkynes .......................................................................................... 41
1.8 Metalloenzymes ................................................................................................................... 43
1.9 Epoxidation .......................................................................................................................... 48
3
1.10 Hydroxylation ...................................................................................................................... 50
1.11 Dissertation summary ......................................................................................................... 52
CHAPTER 2 ......................................................................................................................................... 57
METHODOLOGY ............................................................................................................................... 57
2.1 Solving the Schrodinger equation ......................................................................................... 59
2.2 Born-oppenheimer approximation ........................................................................................ 61
2.3 Slater determinant ................................................................................................................. 63
2.3.1 Anti-symmetry Principle ............................................................................................... 63
2.4 Hartree-Fock theory (HF) ..................................................................................................... 64
2.5 Electron Correlation .............................................................................................................. 66
2.6 Density Functional Theory (DFT) ........................................................................................ 67
2.7 Functionals ............................................................................................................................ 70
2.7.1 Local Density Approximation (LDA) ............................................................................ 70
2.7.2 Generalized Gradient Approximation (GGA) ............................................................... 71
2.7.3 Hybrid Functionals ....................................................................................................... 71
2.8 Basis Sets .............................................................................................................................. 72
2.8.1 Slater-type Orbital (STO) .............................................................................................. 72
2.8.2 Gaussian-type orbital (GTO) ........................................................................................ 73
2.8.3 Polarisation and Diffuse basis functions ...................................................................... 75
2.9 Zero-Point Energy ................................................................................................................. 76
2.10 Transition states .................................................................................................................... 77
2.11 Computational Software ....................................................................................................... 78
CHAPTER 3 ......................................................................................................................................... 79
ABSTRACT ......................................................................................................................................... 80
3.1 Introduction ........................................................................................................................... 81
3.2 Methods................................................................................................................................. 85
3.3 Results and Discussion ......................................................................................................... 87
3.4 Conclusion ............................................................................................................................ 99
4
ABSTRACT ....................................................................................................................................... 101
4.1 Introduction ......................................................................................................................... 102
4.2 Methods............................................................................................................................... 106
4.2.4 COMPUTATION ............................................................................................................. 106
4.3 Results ................................................................................................................................. 108
4.4 Theoretically derived reaction paths, energetics and structures. ......................................... 111
4.5 Discussion ........................................................................................................................... 119
4.6 Conclusion .......................................................................................................................... 128
CHAPTER 5 ....................................................................................................................................... 129
ABSTRACT ....................................................................................................................................... 130
5.1 Introduction. ........................................................................................................................ 131
5.2 Methods............................................................................................................................... 134
5.3 Results ................................................................................................................................. 136
5.4 Discussion ........................................................................................................................... 149
5.5 Conclusion .......................................................................................................................... 161
CHAPTER 6 ....................................................................................................................................... 162
ABSTRACT ....................................................................................................................................... 163
6.1 Introduction ......................................................................................................................... 164
6.2 Methods............................................................................................................................... 169
6.2.1. DFT model calculations .............................................................................................. 169
6.2.2. QM/MM calculations .................................................................................................. 170
6.3 Results and Discussion ....................................................................................................... 173
6.3.1. DFT model reactions .................................................................................................. 173
6.3.3. QM/MM studies of desaturation ................................................................................. 180
6.3.4. Valence Bond rationalization of rebound and desaturation mechanisms ................... 184
6.3.5. Molecular orbital rationalization of rebound and desaturation mechanisms ............ 188
6.4 Conclusion .......................................................................................................................... 191
CHAPTER 7 ....................................................................................................................................... 192
5
ABSTRACT ....................................................................................................................................... 193
7.1 Introduction ......................................................................................................................... 194
7.2 Methods............................................................................................................................... 198
7.3 Results and Discussion ....................................................................................................... 201
7.4 Conclusions ......................................................................................................................... 218
ABSTRACT ....................................................................................................................................... 220
8.1 Introduction ......................................................................................................................... 221
8.2 Methods............................................................................................................................... 226
8.3 Results ................................................................................................................................. 227
8.4 Discussion ........................................................................................................................... 238
8.5 Conclusion .......................................................................................................................... 248
9.1 Concluding remarks ............................................................................................................. 249
References .......................................................................................................................................... 251
LIST OF FIGURES
Figure 1.1: Effect of ligands and valence states on the selectivity in a metal centred catalysed
reaction of butadiene. ............................................................................................................... 22
Figure 1.2: PyBox Ligand from an oxazoline group. .............................................................. 30
Figure 1.3: Example of a typical carbene Ligand .................................................................... 31
Figure 1.4: Dissociative and associative ligand exchange ....................................................... 34
Figure 1.5: insertion mechanism .............................................................................................. 35
Figure 1.6: migration mechanism ............................................................................................ 36
Figure 1.7: disproval of “outersphere” insertion...................................................................... 37
Figure 1.8: charge distribution in a migration reaction............................................................ 38
Figure 1.9: β – hydride elimination.......................................................................................... 39
6
Figure 1.10: Migratory de - insertion ....................................................................................... 40
Figure 1.11: Nucleophilic attack to a coordinated alkene ........................................................ 42
Figure 1.12: Substitution of metal in cofactor to incorporate alternative metal into a binding
site. The Apo-protein must be synthesised first before introducing the new metal. Picture is
obtained from protein data bank 1N2C (Schindelin, H.; Kisker, C. et al 1997). ..................... 44
Figure 1.13: Schematic representation of the catalytic cycle of a typical substrate
hydroxylation reaction of substrate R–H by P450 enzymes and a visual depiction of the active
species (Highlighted in a rectangular box) (Nam, W., et al., 2000). ...................................... 46
Figure 1.14: Reactivity patterns of P40 Compound I with selected substrates. ...................... 48
Figure 1.15: Potential energy profile for the epoxidation reaction of 2-butene by 4,2
[Fe(IV)-
oxo Porphyrin with a thiolate axial ligand as calculated with DFT methods. Data taken from
ref. (de Visser 2012).Values in Kcal/mol. ............................................................................... 49
Figure 1.16: High-lying occupied and low lying virtual orbitals of Compound I at an
optimised geometry of P450 Cpd I of quartet and doublet spin electron arrangements. ......... 51
Figure 2.1: A Potential Energy Surface (PES) indicating a reaction from reactant to product
that passes through a transition state (TS). .............................................................................. 78
Figure 3.1: Relative spin state energies of (5/2,
5/2), (
5/2,
3/2) and (
3/2,
3/2) states of structure 1b
as calculated with various DFT/BS2 methods in Gaussian. Calculations done with basis set
BS3 labeled with superscript a. (a) E values relative to the S = (5/2,
5/2) state. (b) E+ZPE
values relative to the S = (5/2,
5/2) state. ................................................................................... 89
Figure 3.2: Quartet/sextet spin state energies for (A) individual core I and (B) core II. All
energies are relative to the quartet spin state in kcal mol–1
. A negative value implies an S = 5/2
ground state. ............................................................................................................................. 92
Figure 3.3: Optimized geometries of the S = (5/2,
5/2), [S = (
5/2,
3/2)] and {S = (
3/2,
3/2)} states
of 1a•X and 1b•X with X = I3–, BF4
– or ClO4
– at the B3LYP/BS1 level of theory in Jaguar.
Bond lengths are given in angstroms and is the average deviation of the iron from the plane
through the four nitrogen atoms. .............................................................................................. 95
7
Figure 4.1 Active site of P450 as taken from the 2WM4 pdb file. ........................................ 104
Figure 4.2: Time dependence of relative ion abundancies for the reaction of
[FeIV
(O)(TPFPP+•
)]+ (m/z 1044) with indene. Product ions are [Fe
III(TPFPP)]
+ (m/z 1028),
[Fe(TPFPP)(C9H8)O]+ (m/z 1160) and C9H8
+• (m/z 116). Experiments were performed in the
presence of indene at 5.2 10–8
mbar in the FT-ICR cell. .................................................... 110
Figure 4.3: Molecular valence orbitals of 4A. ........................................................................ 112
Figure 4.4: Optimized geometries of the 4,2
A2u and 4,2
A1u states of 4,2
A as calculated
with UB3LYP/BS1 [UB3LYP/BS2] {UB3LYP-D3/BS2} with Fe–O bond lengths in
angstroms. ............................................................................................................................. 116
Figure 4.5: UB3LYP/BS1 optimized geometries of epoxidation transition states with
bond lengths in angstroms. ................................................................................................. 118
Figure 4.6: Correlation between experimental and computational barrier heights. ............... 121
Figure 4.7: (a) Correlation between experimentally determined RT ln kexp (for raw data,
see Table 1) versus known ionization energies (IE). (b) Correlation between calculated
epoxidation activation enthalpy (in kcal mol–1
) and experimental ionization energy for
the substrates in Fig 5. ......................................................................................................... 122
Figure 4.8: VB curve crossing diagram for the C–O bond formation step in olefin
epoxidation (R2C=CH2) by [FeIV
(O)(TPFPP+•
)]+. Valence electrons are identified with
a dot and lines (curved and straight) in the VB structures represent bonds. ................ 124
Figure 4.9: Correlation between calculated epoxidation activation enthalpy (in kcal
mol–1
) and BDECH for the substrates. ................................................................................ 127
Figure 5.1: Oxidants and substrates used in this work. .......................................................... 134
Figure 5.2: UB3LYP/BS1 optimized geometries of 4,2
1X and 4,2
2X in the gas phase with bond
lengths in angstroms. Group spin densities are obtained at UB3LYP/BS2//UB3LYP/BS1 and
are reported in atomic units. ................................................................................................... 138
Figure 5.3: Potential energy profile of styrene epoxidation by 4,2
1Cl as calculated with
UB3LYP/BS2//UB3LYP/BS1. All energies are in kcal mol–1
relative to isolated reactants in
the doublet spin state and include ZPE corrections. Also shown are optimized geometries of
critical points with bond lengths in angstroms and the imaginary frequency in the transition
8
state in wave numbers. Free energies are given in parenthesis and include UB3LYP/BS2
energies corrected with thermal and entropic corrections at 298 K. Data in square brackets
obtained after a UB3LYP-D/BS1 geometry optimization. .................................................... 140
Figure 5.4: Optimized geometries of rate determining transition states 2TSX,Z for the reaction
of para-Z-styrene with 21X (X = Cl
–/NCCH3). Geometries optimized at UB3LYP/BS1 with
bond lengths given in angstroms and the imaginary frequency in wave numbers. Also given
are barrier heights (E+ZPE) for 2TSX,Z with energies calculated at
UB3LYP/BS2//UB3LYP/BS1+ZPE relative to isolated reactants in kcal mol–1
and free
energies of activation in solvent (G+Esolv+Edisp) relative to 2RCX,Z. ................................... 142
Figure 5.5: Analysis of structural features of the transition states TSX,Z calculated at
UB3LYP/BS1 as a function of the height of the epoxidation barrier with respect to the: (a) C–
O distance, (b) Fe–X distance, and (c) imaginary frequency in the transition state. Data given
for X = Cl– axial ligand (diamonds) and X = acetonitrile (squares). ..................................... 145
Figure 5.6: Optimized geometries of epoxidation transition states 2TS′X,Z for the reaction of
2X (X = Cl–/NCCH3) with para-Z-styrene. Bond lengths are in angstroms and the value of the
imaginary frequency in wave numbers. ................................................................................. 147
Figure 5.7: UB3LYP/BS2//UB3LYP/BS1 calculated group spin density ranges for para-Z-
styrene epoxidation by 1Cl (top) and 1AN (bottom). ............................................................... 150
Figure 5.8: Styrene epoxidation barrier heights (E‡+ZPE) of
2TSX,Z plotted against the
ionization energy of the corresponding substrate. (a) Energies relative to isolated reactants.
(b) Energies relative to a reactant complex (RC)................................................................... 153
Figure 5.9: Valence bond curve crossing diagram for para-Z-styrene epoxidation by iron(IV)-
oxo porphyrin cation radical oxidants. Lewis structures give relevant valence-orbitals with
a dot. ....................................................................................................................................... 154
Figure 5.10: Correlations of (a) Epoxidation barrier height of para-NO2-styrene with BDEOH.
(b) Epoxidation barrier height of para-N(CH3)2-styrene with EACpdI(X). .............................. 158
Figure 5.11: Correlation between epoxidation barrier height of all data for 1X (X = Cl–,
NCCH3) with parameter .................................................................................................... 160
Figure 6.1: Extract of the active site of P4502C19 as taken from the 4GQS pdb file. .......... 166
9
Figure 6.2: Potential energy profile of ethylcarbamate activation by 4,2
Cpd I of P450 as
calculated with DFT. Energies are given in kcal mol–1
and are calculated at
UB3LYP/BSII//UB3LYP/BSI level of theory with ZPE and solvent corrections included.
Values in parenthesis are free energies in solvent. Optimized geometries give bond lengths in
angstroms and the imaginary frequency in the transition states in wave numbers. ............... 174
Figure 6.3: Optimized geometries of hydrogen atom abstraction transition states of VA, ET
and DHA by 4,2
Cpd I of P450 as calculated with DFT. Bond lengths are in angstroms, the
imaginary frequency is in wave numbers and (free) energies are given in kcal mol–1
and are
calculated at UB3LYP/BSII//UB3LYP/BSI level of theory with ZPE and solvent corrections
included. ................................................................................................................................. 177
Figure 6.4: Potential energy profile of eicosanoic acid activation by 2Cpd I of P450 as
calculated with QM/MM. Energies are given in kcal mol–1
and are calculated at
UB3LYP/BSIV//UB3LYP/BSIII level of theory with ZPE corrections included. Optimized
geometries give bond lengths in angstroms and angles in degrees. ....................................... 180
Figure 6.5: Geometry scans for the rotation along the Fe–O bond from 2I as calculated
with QM/MM. Energies are given in kcal mol–1
and each data point represents a full
geometry optimization with fixed H–O–Fe–Nheme dihedral angle. Also shown are the maxima
of the scans with key hydrogen bonding interactions identified. The atom labelled with a
yellow star is C. .................................................................................................................... 182
Figure 6.6: Valence bond curve crossing diagrams for product formation from radical
intermediates. (a) Radical rebound leading to hydroxylation products. (b) Hydrogen atom
transfer to give desaturation products. ................................................................................... 184
Figure 7.1: Optimized geometry of 3 as calculated with B3LYP. Bond lengths are given in
angstroms and group NBO charges Q in atomic units. The right-hand-side displays the
natural bond orbitals and their ordering for those involving the central carbon atom with it
ligands. ................................................................................................................................... 202
Figure 7.2: Free energy profile of alkyl chain growth on 3 via either (i) CH3-transfer followed
by H-transfer (mechanism from the center to the left) or (ii) H-transfer followed by CH3-
transfer (mechanism from the center to the right). Free energies (in kcal mol–1
) are obtained
10
with B3LYP-D3/BS2 and contain zero-point, thermal and entropic corrections at 298K.
Values in parenthesis include solvent corrections to the free energy. Optimized geometries
report bond lengths in angstroms, angles in degrees and the imaginary frequency in the
transition states in wave numbers. ......................................................................................... 204
Figure 7.3: Energies of initial H-atom or CH3-transfer reactions from 3 as calculated with
different DFT methods. All structures optimized at B3LYP/BS1 and single point calculations
with basis set BS2 applied. Relative energies are given in kcal mol–1
and include ZPE
corrections calculated at B3LYP/BS1. .................................................................................. 207
Figure 7.4: Optimized geometries of 3’, IMe’ and IH’ with bond lengths in angstroms. ........ 209
Figure 7.5: Valence bond curve crossing diagrams for methyl transfer (part a) and hydrogen
atom transfer (part b) from 3. Valence electrons are identified with a dot. ........................... 211
Figure 7.6: Bond dissociation free energies (BDFEs in kcal mol–1
) of key bonds in structures
3, IMe and IH. Reactions calculated according to Eqs 1 – 4. Part (a) gives adiabatic BDFE
values and part (b) diabatic BDFE values. Values in parenthesis are solvent corrected free
energies, whereas those out of parenthesis are gas-phase data. ............................................. 213
Figure 7.7: Energy decomposition of the methyl and hydrogen atom transfer reaction from 3.
Free energies given are in kcal mol–1
. .................................................................................... 215
Figure 8.1: Active site structure of P450 with key amino acids and substrate (camphor) and
solvent water (W) highlighted. Amino acids labelled as in the pdb file. ............................... 223
Figure 8.2: High-lying occupied and low-lying virtual orbitals of 4RMn. Orbital energies are
reported in au. ........................................................................................................................ 228
Figure 8.3: Optimized geometries with bond lengths in angstroms of 4RMn and
5RFe; group
spin densities () taken from UB3LYP/BS2 calculations. Also given are relative energies
(including ZPE and solvent corrections in kcal mol–1
) of all low lying spin states for RMn and
RFe. ......................................................................................................................................... 231
Figure 8.4: (a) Potential energy profile of hydrogen atom abstraction from DHA by 4,6,2
RMn as
calculated using DFT methods with energies in kcal mol–1
relative to the quartet spin reactant
complex. Energies are taken from the UB3LYP/BS2 calculations in the gas-phase, while
solvent corrected values are in parenthesis. Free energies with solvent, entropic and
11
dispersion corrections are given in square brackets. (b) Optimized geometries of the transition
states for hydrogen atom abstraction from DHA and CHD with bond lengths in angstroms
and the imaginary frequency in the transition state in cm–1
. .................................................. 234
Figure 8.5: Optimized geometries of 4RX structures with different substituents X with bond
lengths in angstroms and the vibrational frequency in wave numbers. Note that 4RMn has X =
t-Bu. ....................................................................................................................................... 236
Figure 8.6: (a) Barrier heights (E‡+ZPE in solvent) for H-atom abstraction from DHA by
various complexes manganese based complexes RX (X = t-Bu, i-Pr, Et, Me or H). (b) Barrier
heights as a function of the Mn–O frequency (MnO) in the reactant complex. ..................... 237
Figure 8.7: Extracts of the active site environments of nonheme iron dioxygenases
representing from left-to-right: TauD (1OS7 pdb), AlkB (3I2O pdb) and CDO (2IC1 pdb).
Amino acids are labelled as in the pdb file. ........................................................................... 243
Figure 8.8: (a) Orientation of substrate attack on the metal(IV)-oxo group with angles in
degrees and group spin densities in au. (b) Electron transfer processes and LUMO orbital that
is filled with one electron in the H-abstraction process. ........................................................ 243
LIST OF SCHEMES
Scheme 3.1: Structures investigated in this work. ................................................................... 84
Scheme 4.1: Models investigated in this work. ..................................................................... 106
Scheme 4.2: Substrates investigated in this work. ................................................................. 108
Scheme 4.3: Pathways observed for the reaction of [FeIV
(O)(TPFPP+•
)]+ ions (R = C6F5) with
selected substrates (Sub) as studied with FT-ICR MS. ......................................................... 110
Scheme 6.1: (a) Competitive hydroxylation and desaturation metabolism pathways of drug
molecules by P450 Cpd I. (b) Reaction products observed for valproic acid and
ethylcarbamate. ...................................................................................................................... 168
Scheme 6.2: Atoms in the QM region of the QM/MM calculation. Wiggly lines represent the
cuts between the QM and MM regions. ................................................................................. 173
Scheme 6.3: Reaction Mechanism of Ethyl Carbamate Activation by Cpd I of P450. ......... 173
12
Scheme 6.4: Orbital mixing patterns for the pathways from radical intermediates to products.
................................................................................................................................................ 189
Scheme 7.1: Catalyst investigated in this work for alkyl formation on a carbide center. ..... 196
Scheme 7.2: Hybridization scheme of 3. ............................................................................... 203
Scheme 7.3: VB description of second reaction steps. (a) hydrogen-atom abstraction. (b)
methyl-transfer. ...................................................................................................................... 216
Scheme 8.1: Models studied in this work. ............................................................................. 225
13
LIST OF TABLES
Table 3.1. Relative energies of optimized geometries in different spin states of 1a•X and
1b•X complexes.a ..................................................................................................................... 96
Table 4.1 Kinetic data and product distributions obtained fro the gas phase reaction of
[FeIV
(O)(TPFPP+•
)]+ with selected olefins as determined by FT-ICR
MS……………………......109
Table 4.2. Relative energies of several low-lying electronic states of [Fe(O)(Por+•)]+
(A).a.......................................................................................................................................113
Table 5.1. Substrate chemical properties and charge-transfer (QCT) values in the transition
states. ...................................................................................................................................... 160
Table 6.1. Free energies of activation of hydrogen atom abstraction, rebound and desaturation
barriers.a ................................................................................................................................. 178
Table 8.1. Calculated barrier heights for hydrogen atom abstraction from dehydroanthracene
and 1,4-cyclohexadiene by metal-oxo complexes (Energies are in kcal mol–1
). ................... 244
WORD COUNT: 69,110
14
ACKNOWLEDGEMENT
My deepest appreciation goes to my distinguish supervisor Samuël P. de Visser who possess
the manner, support and the essence of an intellect: he persistently conveyed strength of
adventure in consideration to research and brotherly support; without his guidance and
persistent help this research would not have been possible.
I am heartily thankful to Mr. Balarabe Z. Ahmed for giving me the moral support I require.
I also thank the National Service of computational chemistry software (NSCCS) for a
generous CPU time.
I cannot exempt my appreciation from my research colleagues within the MIB building and
the entire University faculty and the support staff for their help and patience.
And finally I would also like to take this opportunity to thank Petroleum Technology
Development Fund (PTDF) for its generosity in funding my study; I am very honoured to be
a recipient of this award.
15
ABSTRACT
The University of Manchester,
School of Chemical Engineering and Analytical Science
ABSTRACT OF THESIS submitted by Mala Alhaji Sainna for the degree of Doctor of
Philosophy (PhD) and entitled “Electronic and catalytic properties of iron porphyrin
complexes: Trends and reaction mechanisms”
The cytochrome P450s belong to the superfamily of proteins containing a heme cofactor and,
thus, are termed hemoproteins. They perform important oxidation reactions in the body, and
are, for instance, involved in the metabolism of drug molecules in the liver as well as the
detoxification of xenobiotics and biosynthesis of hormones. The active species of these
enzymes is identified as iron(IV)-oxo heme cation radical species (also known as Compound
I), but it is short-lived and difficult to trap and characterize experimentally. Therefore,
theoretical modelling as implemented in this thesis was instituted in order to give important
answers to questions such as the mechanisms of substrate monoxygenation, the oxidant
activity in enzymes and the effect of protein architecture on chemical catalysis. The present
thesis focuses on addressing these issues using a combination of density functional theory
and quantum mechanics/molecular mechanics studies. The work gives insight into the nature
of heme, axial ligand bound to heme and the impact of substrate on oxidation reactions. We
find trends in reaction mechanisms and rate constants and rationalize rate constants and
reaction processes. The work has also given insight into the nature of high valent iron(IV)-
oxo heme cation radical oxidants and their reactivity patterns with respect to a broad range of
substrates. Hence, the offered studies have shown how small structural differences in the
active site will result in dramatic differences in reactivity patterns and how nature approach
and catalyzes vital reaction mechanisms.
16
DECLARATION
No portion of this work referred to within the thesis has been submitted in support of an
application for another degree or qualification of this or any other university or institute of
learning.
17
COPY RIGHT STATEMENT
i. The author of this dissertation including any appendices and/or schedules to this dissertation
owns certain copyright or related rights in it and s/he has given the University of Manchester
certain rights to use such copyright, including for administrative purposes.
ii. Copies of this dissertation, either in full or in extracts and whether in hard or electronic copy,
may be made only in accordance with the copyright, Designs and Patents Act 1998 (as
amended) and regulations issued under it or, where appropriate, in accordance with licensing
agreements which the university has entered into. This page must form part of any such
copies made.
iii. The ownership of certain copyright, patents, designs, trade marks and other intellectual
property (the “Intellectual Property”) and any reproductions of copyright works in the
dissertation, for example graphs and tables (“Reproductions”), which may be described in
this dissertation, may not be owned by the author and may be owned by third parties. Such
intellectual properties and reproductions cannot and must not be made available for use
without the prior written permission of the owner(s) of the relevant intellectual property
and/or reproductions.
iv. Further information on the conditions under which disclosure, publication and
commercialisation of this dissertation, the copyright and any intellectual property and/or
reproductions described in it may take place is available in the University IP Policy (see
http://documents.manchester.ac.uk/display.aspx?DocID=487), in nay relevant Dissertation
restriction declarations deposited in the University Library, The University Library
regulations (see http://www.manchester.ac.uk/library/aboutus/regulations) and in the
University’s Guidance for the Presentation of Dissertations.
18
CHAPTER 1
INTRODUCTION
19
Introduction
1.1 Preface
Catalysis can be classified into homogeneous catalysis in which the reactant and the catalyst
exist in the same phase, and a heterogeneous catalysis in which the reaction occurs on the
interface between different phases; for instance the substrate is in solution and reacts with a
solid catalyst surface. Nowadays transition metal catalysts are the most commonly applied
examples of both homogeneous and heterogeneous catalysts (Busca, G.,2014). Undoubtedly,
although some of the inner transition metals, such as Pd, Ru, Ir and Rh may be efficient
catalysts for chemical reactions, it may not always be practical in an industrial setting to
utilise those catalysts due to difficulties to obtain and purification. Thus, their toxicity and
reactivity in nature leads to environmental and economic problems associated with their
usage. Consequently, industry is always seeking to find more environmentally friendly
alternatives for process technologies (Reetz, M.T.,2014). Therefore, current drives in
chemical research is replacement of toxic and expensive second and third row catalysts with
environmentally benign and cheaper options, and thus work has focus on developing catalysts
that use iron instead (Enthaler, S.,2013).
Transition metal catalysis has revolutionised organic chemistry enabling chemical
transformation of molecules that is very difficult to achieve otherwise (Thomas, J.M.,2014).
Some of the uniqueness of transition metals is their ability to offer varieties of oxidation
states and coordination spheres, which makes them highly susceptible to electron abstraction
and/or donation processes. The nature by which these processes act can have a strong
influence on, the regioselectivity as well as the stereospecificity of chemical reaction,etc
20
(Kratsch, J., Roesky, P.W.;2014). Moreover, the ability of transition metals to bind multiple
atoms of a single ligand (hapticity) leads to interesting binding patterns in organometallic
complexes (de Visser, S.P.,2001, Shelby, M.L., Mara, M.W. et al 2014). In nature, these
transition metal catalysts form the class of enzymes called metalloenzymes, which are highly
multipurpose and take part in biological processes that include, for instance, metabolism,
biosynthesis and biodegradation processes. Often, these metalloenzymes utilise iron in their
active centre bound to the protein via a multidentate ligand system containing a heme or a
non-heme structure. One class of the heme metalloenzymes that are extensively studied are
the cytochrome P450s, which are monoxygenases, For example, The P450s are involved in
drug metabolism reactions in the liver (Gonzalez-Diaz, H., 2013, de Visser, S.P. 2013). These
catalysts have an extensive array of reactivity patterns; however, all proceed via short-lived
intermediates, which are difficult to study experimentally. With the advance in computational
methods, we are now able to investigate the behaviour and the electronic properties of these
species in detail, and are also able to predict the spectroscopic parameters of short-lived
intermediates (Shi, S.; Klotz, U. 2012, Murphy, C., 2014, Nyman, G., 2014, Tsipis, A.C.,
2014, Fioroni, M., Dworeck, T. et al 2014).
In this thesis, we discuss the reactivity patterns and chemical properties of catalytic cycle
intermediates of enzymes and synthetic model complexes. Our studies give insight into the
nature and the effect of ligands in the reaction mechanism, the rate constant and the product
distributions. Moreover, our work investigated the influence of the environment of the
reactants as well as their activity towards substrates.
21
1.2 Homogeneous Catalysis
Homogeneous catalysis utilising transition metal complexes is one of the most studied
research areas and has received an enormous boost in recent years. Many interesting
discoveries were made over the years both by researchers and industries as well as in
academia. The homogeneous catalysis can be defined as a catalytic system in which the
reactant or the substrates for a reaction and the parent catalyst component are all merged
within a single phase, usually the liquid phase. Lately, a tapered definition has become viral
according to which homogeneous catalyst involves organometallic complexes as the
catalysts; although, it is a clear fact that an organometallic compound has a bond between the
central metal and a carbon atom which is not the case with all known homogeneous catalysts.
Therefore, an organometallic complexes can be categorised as a class of homogeneous
catalyst but there are few more interesting and important reactions in the literature that
employ homogeneous catalyst which are distinct from organometallic complexes based on
their ligand systems (Anderson, G. K.;1984, Mawby, R.J.; Basolo F.; et al 1964, Noack,
K.;Calderazzo, F. 1967). Examples of some of these catalysts include:
Diels-Alder reactions (Lewis acid as catalysts),
Ester hydrolysis which is a General acid and base catalysis,
Organic catalysts (thiazolium ion in Cannizzarro reactions),
Co-ordination complexes,
Enzymatic processes,
Porphyrin complexes,
Biomimetic complexes
22
Although the above listed catalysts are categorised as homogeneous catalysts, they primarily
differ by their ligands effect and coordination. A single metal can give a variety of products
from the same substrate by simply altering its ligand environment. One good example found
in literature relates to the varieties of substrates formed by a single metal by simply changing
the ligands around the metal centre as indicated in Figure 1.1. The figure shows the various
products obtained from 1, 3-butadiene using a nickel catalyst with different ligand
coordination.
Figure 1.1: Effect of ligands and valence states on the selectivity in a metal centred catalysed
reaction of butadiene.
We will not discuss every class of homogeneous catalyst in detail in this chapter; however we
are going to give a brief description of the few we can relate to within our work.
23
1.2.1 Ligand effects.
Homogeneous catalysis usage represents an important and resourceful method for carrying
out chemical transformations. There are many processes in industry that utilise organo-
transition metal catalysts (Flood, T.C.; Jensen, J.E.; et al 1981, Kataoka, Y.; Shibahara, A.; et
al 2001, van Leeuwen, P. W.N.M; Roobeek, C.F. et al 1994). The development of selective
chemical processes generally requires modifying ligands on the transition metal centre. These
ligands have a unique ability to stabilise metals in several oxidation states and geometries,
and they can also be tuned to radically change the reactivity of a catalyst. Hence, ligand
systems are one of the several tools utilised by chemists in studying and subsequently
controlling the reactivity of homogeneous catalysts. Apart from the ligand-based approach in
the control of transition metal reactivity, there are alternative approaches, including those
focused on altering the d-electron energy of the metal through either metal based oxidation or
reduction. This procedure can help in controlling the electronic and steric properties of the
metal coordination sphere, although the ligand-based approach is more widely used. These
procedures can also be used in controlling the thermodynamic binding affinities of transition
metals, for instance, it has been shown that reactivity pattern can be changed of a centrally
bounded transition metal as a function of ligand oxidation state as a result of redox-active
ligands. In particular, these redox-active ligands enable the stability of different oxidation
states and, thereby, resulted in transition metals that produced compounds with switchable
electronic state of reactivity (van Leeuwen, P. W. N. M.; Morokuma, K.; et al 1995, Maseras,
F.; Lledós, A. 2002, Dekker, G. P. C. M.; Buijs, A.; et al 1992). Moreover, it was found that
the catalyst produced from these kinds of ligands could be a sluggish catalyst in one
particular oxidation state and brilliantly active after single or multiple oxidation or
24
reduction.Furthermore, they were also found to display selectivity for a definite
transformation depending on the oxidation state associated with the complex (Lundquist, E.
G.; Folting, K.; et al 1990). The redox-active ligands can also be used as an electron shuttle
in a catalytic process as shown by Hembre and McQueen (García Alonso, F. J.; Llamazares,
A.; et al 1991).
1.2.2 Bite angle effect
The term bite angle is usually associated to geometric parameters used to classify chelating
ligands in coordination chemistry; in particular, structures involved in organometallic
chemistry and some biomimetic and enzymatic complexes. Although it was commonly
applied to phosphine ligands due to its ability to adopt a wide range of chelating ring sizes,
the parameters are also applicable to any sort of chelating ligands. We are however going to
use the widely understood diphosphines as an example to give a brief explanation of these
features. Many instances indicate that the ligand bite angle is somehow associated to catalytic
performance in a number of reactions. Early examples are the platinum diphosphines
catalysed hydroformulations (Versluis, L.; Ziegler, T.; et al 1990). In recent years, a
connection between the ligand bite angles and the selectivity of a catalyst has been observed;
a specific example is the rhodium catalysed hydroformylation, the Diels-Alder reactions as
well as the nickel catalysed hydrocyanation (Brookhart, M.; Green, M. L. H. 1983). The L-
M-L angle found in transition metal complexes is believed to be a concession between the
attached ligands preferred bite angle and the angle preferred by the metal centre. The former
is mainly determined by constraints forced by the ligand atoms and the backbone. Electronic
effects seem to have a more indirect influence by changing the preferred metal-ligand bond
25
length. The metal preferred bite angle however, is primarily determined by electronic
requirements; for instance the nature and number of d-orbitals associated in the formation of
molecular orbitals. Other ligands attached to the metal centre can also have influence on the
metal orbitals, for example, -bonding ligands.
As per the aim in explaining the effect of bite angles in the catalytic reactions, there are
distinctions worth mentioning between two different effects (Mole, L.; Spencer, J. L.; et al
1991, Conroy-Lewis, F. M.; Mole, L.; et al 1991): (1) steric bite effect: This effect is related
to the steric interactions between ligand-ligand or ligand-substrate, which is observed when
the bite angle is modified by altering the backbone and keeping the attached substituents at
the donor atoms the same. The steric interaction observed from this has an influence upon
altering the energies of the transition states and the catalyst resting states, thus modifying the
activity or selectivity of the main catalytic system. (2) electronic bite angle effect: As the
name implies this is associated to the electronic changes within the catalytic centre as the bite
angle is altered (Thorn D. L.; Hoffmann, R. 1978). It can be defined as an orbital effect, due
to the fact that the bite angle determines metal hybridisation and as a consequence metal
orbital energies and reactivity. This effect can also show itself as a stabilisation or as a
destabilisation at the initial reaction, final or at the transition state of a reaction. It was also
observed that when the substituents at the donor atom are kept the same while the bite angle
is changed, the steric properties also change (Hickey, C. E.; Maitlis, P. M. 1984, Amatore, C.;
Jutand, A. 2000).
26
1.2.3 Cone angle effect
The cone angle notion was found to be of practical importance when dealing with
homogeneous catalysis due to the fact that the size of the ligand could affect the reactivity of
the attached metal centre. A known example shows where the selectivity of hydroformylation
catalysts was strongly influenced by the size of the co-ligands (Calhorda, M. J.; Brown J. M.;
et al 1991, Widenhoefer, R. A.; Buchwald, S. L.; 1998); Thus, some phosphine ligands are
wide enough to conquer more than they should within the coordination sphere of a metal
centre. The concept of cone angle was first introduced by Tolmans parameter Ѳ (theta) in an
attempt to define a reliable steric parameter complementary to the electronic parameter.
Tolman proposed to measure the steric bulk of phosphine ligand from CPK models (CPK
models are also called Space-filling models after the chemists Robert Corey, Linus
Pauling and Walter Koltun, who founded their use) (Robert B.; Pauling, L. 1953). They
constructed a cone which embraces all the atoms of the substituents on the phosphorus atom
at a distance of 2.28Å from the metal centre and the cone angle is then measured which are
considered as the desired steric parameters. Tolman proposed quite a few other methods
which were later modified according to the chemical series investigated. Sterically, it is
understood that more bulky ligands give less stability in complexes. In a series of comparable
ligands, such as aryl phosphites, this leads to a reliable index. Crystal structure determination
has indicated that practically the angles recognized within the structure complexes are smaller
than the Ѳ-vales would propose. For instance, in the case of cis-triphenylphoshine molecules,
a double of the cis-triphenylphoshine may give an angle of 95o between P-M-P, whereas the
Ѳ-values would actually predict 145o in reality interlinking of the R-substituents leads to
smaller effective cone angles. In some instances, the steric interactions nearby the metal
27
centre could be important while for other properties interacting in a more distant from the
metal centre may dominate, but the ligands never form a perfect cone angle (Fitton, P.; Rick,
E. A. 1971, Goertz, W.; Kamer, P. C. J.; et al 2001).
Recently, numerous efforts have started in order to describe the steric properties of ligands
using molecular mechanics and analysis of data taken from X-ray studies as well as data
mining (Marcone, J. E.; Moloy, K. G. 1998). The accessible molecular surface (AMS)
method is used. In the AMS method, the effective contours of the ligands are calculated,
resembling that of the calculation of accessible surface for enzymes. Crystallographic data
have also been utilised to calculate and compare steric ligand properties, more in particular
for bidentate ligands (Brown , J. M.; Cooley, N. A. 1988, Schrock, R. R. 1979).
1.3 Ligand according to donor atoms.
1.3.1 Imido and Alkoxy groups as an anionic ligand.
Alkoxides and imido groups are known anionic ligands used in titanium and zirconium
catalysts for alkenes polymerisations; they are either used in the catalysts as the only anions
or in combination with cyclopentadienyl ligands (Bäckvall, J. E.; Åkermark B.; et al 1979).
Alkoxide ligands are also used in titanium catalysts for epoxidation of alkenes. The
alkoxides when connected to a neutral ligand like an imine or a phosphine are known to be
excellent ligands for divalent metal catalysts of nickel and palladium for alkene reactions, in
which the second valency of the complex is made up by a hydrocarbyl group involved in a
chain growth reaction or a hydride during the chain transfer process. The alkoxides and
amides are involved in hydrogen transfer catalysts where they play an active part in terms of
abstracting a proton from the alcohol substrate (Luo, X.-L.; Crabtree, R. H. 1989); whereas
28
the elements of column 1-3 on the periodic table such as oxygen and nitrogen are much more
stable when bound as an anionic ligand to a metal centre. In the case of transition metal
complexes the differences between oxygen or nitrogen-based anions is much smaller, even
more so as you proceed towards the right hand side of the periodic table (Luo, X.-L.;
Crabtree, R. H. 1989). As a consequence a large number of Metal-to-heteroatom bonds
contribute in catalytic reactions leading to carbon-to-heteroatom bonds and it allows us to
carry out a large part of our organometallic chemistry in the presence of alcohol, amines or
water etc.
1.3.2 Neutral and Anionic hydrocarbyl groups
Aromatics such as the cyclopentadienyl appear as ligands and became extremely important in
catalysis for transition metals such as Ti, Zr, Hf and Ru for many decades. They also occur as
ligands in ruthenium complexes that are used in hydrogen transfer reaction where two
hydrogen atoms are transferred from donor molecules such as alcohol, to a ketone yielding
another alcohol as the product. Emphasis was made on the enantiospecific variant (Vaska, L.
Diluzio, J. W. 1962, Vaska, L. 1968). For example, when a weakly coordinating anionic
ligand is applied to an early-transition metal that is cationic, these metal ions can be solvated
by aromatics if no stronger ligands are present. Aromatics ligands have as well been used in
nickel complexes and shown to make the catalysts highly active in the addition or vinyl-type
polymerisation of norbornenes (James, B. R.; Mahajan, D. et al 1979).
1.3.3 Oxazolines, imines , Amines and related ligands
All the above listed compounds contain a donating nitrogen atom. A nitrogen ligand is known
to be having the strongest donating function and has been used for many years in catalysis.
29
They occur naturally in imidazoles and porphyrins and bind to transition metals easily, and
they are known to be involved in oxidation reactions (Ungváry, F. 1972). Several of these
complexes were mimicked and are used in the field of homogeneous catalysis to understand
reactions such as, the oxidation of C-H bonds or phenol oxidative coupling reactions (Major,
A.; Horváth, I. T.; et al 1988).
The amines and pyridines are among the oldest most studied ligands in the field of
coordination chemistry and catalysis. For instance pyridines were used in hydrogenation
catalysts based on copper (I) and (II) long before phosphines in platinum complexes
(Mirbach, M. F. 1984), and the sp3 hybridised nitrogen in amines are considered an analogue
of phosphines; they are known to be strong -donor capability as compared to phosphines,
and they are excellent in stabilising high-valent metal complexes such as tetravalent
palladium and platinum (Chinn, M. S.; Heinekey, D. M. 1990). Nitrogen ligands are
generally much more stable than phosphines especially in pyridine and imidazole ligand
complexes. Pyridine complexes, for instance, are not susceptible to oxidation as phosphines,
they are also not known to follow the decomposition pathways of phosphines such as C-P
bond cleavage, phosphide formation or the phosphide hydrolysis (Van Leeuwen, P. W. N.
M.; Roobeek, C. F.; 1986), hence nitrogen ligands are by far more preferable than phosphine
in oxidation catalysis. The bonding characteristics of pyridines or any other sp2 hybridised
nitrogen to that of phosphorus donor ligands are quite distinct. Although they are both good
-donor and poor -acceptor ligands, they both stabilise higher oxidation states of metal
complexes instead of a low oxidation states. For instance in the catalysis of divalent metal or
nickel they have a very similar behaviour, however, in the case of reductive elimination
30
reactions leading to zero-valent complexes, they have different characteristics and
performance, whereby phosphorus is more preferable in this instance.
The oxazolines are thought to have an additional advantage in comparison to other imine that
are nitrogen based ligands due to the fact that the asymmetric derivatives are readily
accessible. A very good example of an oxazoline is the group of PyBox ligands (Figure 1.1)
which is a typically chiral oxazolines (Inuki, T.; Kojima, T. 1967). The mono-oxazoline has
also found a wide recognition and application, the mono-oxazoline contain a phosphine group
as a second donor atom. The substitution of one of the oxygen atoms in oxazoline by a
nitrogen atom yield an imidazoline which contain the asymmetric carbon atom but offer an
electronic variation through substitution at the amine-nitrogen atom (Fleming, I. 1976). The
imidazoline substituent might as well be applicable for immobilisation via covalent linkages.
Figure 1.2: PyBox Ligand from an oxazoline group.
1.3.4 Carbon monoxide and Carbene
Carbenes act both as an intermediate and as a ligand in the field of catalysis. They are known
to occur as an intermediate in a reaction of alkene metathesis as well as alkene
cyclopropanation. They carry hydrogen and carbon substituent as intermediates, hence they
31
are classified as a “Schrock carbenes”. As ligand, however, they often contain nitrogen
substituent and are therefore classified as a “Fischer carbenes”. They have received a great
attention within the last few years as ligands in metal complexes catalysis (Denmark, S. E.;
Fu, 2003, Landmana, M.; Pretorius, R.; et al 2014), however the structural motive was
already in existence and in exploration since the early seventies (Ueki, M.; Matsumoto, Y.; et
al 2001, Sainna, M. A., Singh, D. et al 2015).
Figure 1.3: Example of a typical carbene Ligand
The carbine as ligands, are strong -donor and rarely -acceptors, they are sometimes called
“singlet carbenes”.
The carbon monoxides are often used as either a reactant or a ligand. It is found to be a very
strong -acceptor as a ligand and a modest -donor. In comparison of its steric properties to
that of a phosphorus ligand, it is believed to be one of the smallest ligands available with a χ-
value of around 55 when characterised using IR. One of the exciting features of the CO
ligands is its ease of characterising using IR as mentioned earlier and also in situ IR, in a
transparent frequency area (1800-2200 cm-1
). Although its disadvantages include its demand
of high pressure of about 10 bar in order to obtain a concentration of order 0.1M in an
organic liquid.
32
1.4 Common anions
In most cases the anions form part of the coordination complexes as well as the
organometallic complexes and their role should not be undervalued. In many cases they act as
reactants, for instance in cross-coupling reactions whereby a salt is made as a second product
and even in the simple series of halides its anion function toward the metal centre makes an
enormous difference in catalytic behaviour (Ishihara, K.; Kurihara, H.; et al 1998). The
function of halides and carboxylates as a ligand goes beyond doubt as they are known to bind
strongly to metal ions. For instance, a carboxylate bound to a phosphine is a moiety found as
an impressive ligand in the shell process for ethane oligomerisation (Reetz, M. T.; Kyung, S.
H.; et al 1986), whereas an iodide which is a halide family is found to be important ingredient
of methanol carbonylation chemistry as they are known to form part of the ligand
environment in iridium and rhodium metal complexes (Ishihara, K.; Ohara, S.; et al 2000).
The introduction and implementation of weakly and non-coordinating anions (WCA, NCA)
has also received a great attention over the years as they are indispensable in homogeneous
catalysis. The cationic counter-ions obtained as a result of using a WCA is willingly
accessible for the reactants especially reactions involving alkenes such as polymerisations,
carbonylations and hydrogenations reactions. The introduction of weakly coordinating anions
has led to numerous advances. In the early history of homogeneous catalysis the role of
“Ziegler catalysts” where alkene coordination is facilitated by the role of creating vacant sites
at a metal ion was played by Lewis acids abstracting chloride ions from the catalyst precursor
(Garrou, P. E. 1985), but this has changed in the recent years as the WCA has also headed to
33
an additional control over the species formed in the reactor compared to the use of Lewis
acids.
1.5 Vacant site Creation and coordination of the substrate
The sole purpose of a catalyst is to bring the reactant together and alter the activation barrier
of the reaction; however bringing the reactants together requires a metal centre with a vacant
site. Metal catalysis starts with the creation of a vacant site we could say. For a homogeneous
catalyst in a condensed phase creating a vacant site can be challenging as a condensed phase
solvent molecules will always be co-ordinated to the active metal ion and therefore “vacant
site” is an inaccurate description and this is because substrates are present in excess and so
are the ligands. Hence, a competition in complex formation exists between the desired
susbstrate and other potential ligands present within the same phase or the solution. Usually a
negative order in one of the ligands concentrations can be found in the rate of product
formations expression; a zero order in the concentration of the substrate is obtained when the
substrate co-ordinates strongly to the metal centre. i.e. saturation kinetics (Billig, E.;
Jamerson, J. D.; et al 1980), also known as c.f. Michaelis-Menten kinetics (Goel, A. B. 1984).
Another consideration needs to be taken into account when creating a vacant site and co-
ordination of substrate is the classical way by which substitution reactions are described
(Sakakura, T. 1984). Two thrilling mechanisms are distinguished, an associative mechanisms
and a dissociative mechanism. In the case of dissociative mechanism the breaking of the
bond between the metal and the leaving ligand is the rate controlling step, a solvent molecule
thereby occupies the open site which is a phenomenon that does not appear in the rate
equation. Subsequently the replacement of the solvent by the substrate occurs in the first step.
34
In the associative process (SN2) however, the displacement is the bimolecular process with
simultaneous bond breaking of the ligand and the formation of the metal with the substrate
(Abatjoglou, A. G.; Billig, E.; et al 1984); In square planar complexes as found for the group
9 and 10 metals, the associative process is most common.
Figure 1.4: Dissociative and associative ligand exchange
1.6 Insertion versus migration
The terms insertion and migration denotes the process in which an unsaturated molecule
inserts to a metal-anion bond. In Figure 1.4 and 1.5, a typical instance of the aforementioned
reaction showing acetyl fragment as formed from a co-ordinated CO and a methyl group on a
platinum complex, the two reacting groups mentioned must occupy positions cis to one
another otherwise the reaction does not occur (Abatjoglou, A. G.; Bryant, D. R. 1984).
35
Figure 1.5: insertion mechanism
The most important difference between the insertion mechanism and the migration
mechanism is that, the insertion mechanism consist of carbon monoxide inserting into the
metal-methyl bond while the acetyl bond formed takes the place of the methyl group; that
means the -bonded fragment retains its position trans to *, whereas in the migration
mechanism the methyl group moves to the co-ordinated carbon monoxide resulting to an
acetyl group occupying the position cis to *. The mechanisms depicted in Figure 1.5 above
is evidenced by NMR spectroscopic studies on a platinum complex of diphosphine with the
platinum containing slightly distinct phosphine groups such that they can be distinguished in
the NMR spectrum of both the acetyl and alkyl species (Kong, K-C.; Cheng, C-H. 1991). The
example of migration mechanism with rhodium complex above has proven the fact that
migration mechanism is the path way for complexes containing “piano-stool” structures;
theoretical calculation also ascertained the migration mechanism (Kikukawa, K.; Takagi, M.;
et al 1979), showing the anionic methyl group shifting to the positively charged carbon atom.
Therefore the migration is considered more accurate for this process. Although there are
several experiments that suggest the insertion is the actual mechanism. Most of the systems
studied consist of Mn and Fe as the metal involved. A typical example on a platinum
complex is depicted in Figure 1.6 below. The most stable starting geometry indicated that the
methyl group and the phosphine are in cis positions, because of the influence of the trans. The
36
same mechanism happens for the acetyl product. This explanation might be considered as a
proof for an insertion reaction, however, as the figure shows the migration might be followed
by an isomerisation or an isomerisation may take place after the migration. This is reinforced
by the fact that asymmetric ligands usually undergo insertion reaction pathway more slowly
than the symmetric ligands. Thus, diphosphine and bipyridine complexes undergo faster
reactions as compared to the mixed phosphine-nitrogen ligands (Sisak, A.; Ungváry, F.; et al
1983). Generally, the experiments discussed do not disprove the cis-migration mechanisms
though, although it presumably resulted from topomerisations of the intermediates or
products (Bouaoud, S-E.; Braunstein, P.; et al 1986).
Figure 1.6: migration mechanism
According to the depicted insertion reaction mechanisms shown in the Figure 1.4 above, we
assumed the reacting carbon monoxide is co-ordinated to the Pt metal. There is a clear
37
evidence proven experimentally that methyl migration is indeed co-ordinated to carbon
monoxide (Szulc, A.; Meyerstein, D.; et al 1998). The standard evidence comes from a
relatively inert complex where both the migration as well as the exchange of the co-ordinated
CO with free CO is slow. A clear example is The reaction of CH3Mn(CO)5 as shown in the
Figure 1.7 below, result in the formation of CH3(CO)Mn(CO)4(13
CO) when reacted with 13
C
labelling free CO in which the CO is present as co-ordinating carbon monoxide instead of the
acetyl group (Alcock, N.W.; Bergamini, P.; et al 1987). Therefore, there does not seem to be
a direct evidence of reaction between the methyl manganese unit and the freshly received
carbon monoxide. There is no established proof an insertion of an uncomplexed unsaturated
susbstrate into a metal-to-carbon -bond. It is worth mentioning as in when dealing with
heterogeneous catalysis, however, coordination of the susbtrate to an active metal surface is
acknowledged as the “Langmuir-Hinshelwood” mechanism, whereas the reaction of a
molecule in a gas-phase with some of its fragments on the surface is named the “Eley-Rideal”
mechanism.
Figure 1.7: disproval of “outersphere” insertion
Another essential type of the migration reaction mechanism is the one consisting of alkenes
in place of carbon monoxide as discussed earlier. In this type of migration reaction a hydride
migrates to a co-ordinated ethene molecule cis to the hydride which may result in an empty
space in the coordination sphere of the metal. There are two ways of lifting these sort of co-
38
ordinative unsaturations: (1) either an agostic interaction of the unsaturated metal with the -
hydrogen may occur which is a mechanism supported by both experimentalists and the
theoretician (van Leeuwen, P. W. N. M.; Roobeek, C. F.; et al 1990); or, (2) the occupation
of the vacant site by an incoming ligand.
The activation of a co-ordinated alkenes taking place before migration is currently unclear
although the coordinated alkenes are subject to -back donation and-donation but the
overall result of the electron density is unpredictable. Although, molecular orbital
calculations at an extended level using Hückel theory (Goodson, F. E.; Wallow, T. I.; et al
1997) indicated that in many instances the co-ordinated alkenes are not activated towards
nucleophilic attack (migration) and cannot a priori be that the hydride will undergo a rapid
migration. A strong back-donation leads to more electron-rich alkenes which reduce their
vulnerability for the attack of the migrating group. A strong polarisation of the alkene will
occur when an asymmetric bonding of the alkene is invoked as shown in Figure 1.8 below.
The depicted structure representative below, the alkene is shown to be activated towards the
hydride migration (Bianchini, C.; Meli, A. 1998, Chen, J.B.; Angelici, R. 2000).
Figure 1.8: charge distribution in a migration reaction
39
1.6.1 β-Elimination and de-insertion:
The β-elimination literally refers to the reverse of the migration of η-bonded anionic groups
to co-ordinated alkenes. The migration reaction (or nucleophilic reaction) reduces the total
number of electrons of the complex by two, thereby creating a vacant site at the metal,
whereas the -elimination behaves the exact opposite as it requires a vacant site at the
complex (abandoning coordinated solvents in the complex), and during the reaction process
the electron count is increased by two instead. Reactions resembling the -elimination occur
in many organic reactions although the variation is in the intramolecular nature of the present
process as the eliminated alkene may be retained within the complex. In organic chemistry, a
two-step process is observed in the reaction mechanism, for instance proton elimination with
a base followed by the leaving of the anion. In transition metal chemistry, however, the
availability of d-orbitals facilitates a concerted cis -elimination (Dong, L.; Duckett, S. B.; et
al 1992).
Figure 1.9: β – hydride elimination
The -elimination is a reaction that is often required to be suppressed in order to achieve a
desirable feature. There are a few ways to achieve the suppression of the -elimination:
40
1. By maintaining a coordinative saturation of the complex although this might be a
counterproductive idea as the next step in the catalytic cycle of the complex will
surely require a vacant site to react.
2. By the selection of metals that are stable with metal alkyl complexes with esteem to
hydride and liberated alkene. For the metals on the left side of the periodic table such
as the early transition metals and the lanthanides, there is a relative stability with the
alkyl, hence the best alkene polymerisation catalysts are obtained among these
aforementioned metals.
3. Steric hindrance could be instigated in order to avoid the correct stereochemistry
required for -elimination leading to stability of the metal alkyl complex. In the
modern polymerisation catalysts for polypropene this feature is observed, thereby
leading to polymers with higher molecular weight.
The de-insertion is also used instead of the -insertion especially when dealing with CO. The
process is absolutely similar because it also requires a vacant site for the reaction to occur,
and the electron count of the metal rises by two in the process of de-insertion. In the de-
insertion, an insertion takes place between a -bonded and -bonded fragment in mutual cis-
positions, as was discussed earlier. The mechanisms of de-insertion only proceed if there is a
vacant site cis to the acyl group. A typical example of the de-insertion is shown in the Figure
1.10 below as provided by the experimentalist (Kaneda, K.; Sano, K.; et al 1979).
Figure 1.10: Migratory de - insertion
41
1.6.2 Oxidative addition and Reductive elimination:
In an oxidative addition reaction, a compound AB is bound to a metal complex M during
which the bond of the bounded compound AB is broken to form a new bond with the metal
MA, and MB.A while B is reduced, and both will obtain a minus one charge thereby raising
the oxidation state of the metal by two, also the co-ordination number of the metal also
increases by two, while the electron around the metal complex raises by two and the d-
electron count of the metal decreases by two electrons. In oxidative addition, electronic
ligand effects are exceedingly likely. Donors strongly stimulate the formation of high-
valence states and therefore give rise to oxidative additions.
Reductive elimination is considered the reverse reaction of the oxidative addition discussed.
In the reductive elimination the formal valence state of the metal is reduced by either two
electron or one for a bimetallic reaction thereby reducing the total electron count of the
complex by two. Stabilisation of the low-valent state of the product promotes reductive
elimination; this means a good -acceptor ligands, bulky ligands as well as ligands preferring
bite angles more suited for tetrahedral than for square-planar complexes, when dealing with
elements of group 10 metals.
1.7 Activation of a substrate toward nucleophilic attack
1.7.1 Role of Alkenes and alkynes
Bounding of an alkene to an electronegative metal or often a positively charged metal tend to
activate the alkene toward a nuclophilic attack. Once the nucleophilic attack occurs, the
alkene complex will convert into a -bonded alkyl complex with the nucleophile at the β-
42
position (Chan, A. S. C.; Caroll, W. E.; et al 1983). The overall product obtained from the
nucleophilic attack is virtually the same as that of insertion reaction mechanism; the two are
distinct by the fact that the insertion reaction gives rise to a syn-addition while the
nucleophilic attack to an anti-addition. Sometimes the two reaction mechanisms are referred
to as inner and outer sphere attack for the insertion and nucleophilic attack reactions
respectively. Properly substituted alkenes could demonstrate either the syn or anti manner of
the addition reaction. These types of addition reaction is known as a key-step in palladium
catalysed reactions in “Wacker-type” process (van Leeuwen, P. W. N. M. 2001, Chen, J.;
Daniels, L. M.; et al 1990, Angelici, R. 1988).
Figure 1.11: Nucleophilic attack to a coordinated alkene
Although, the activation of the alkene by the metal toward nucleophilic attack is unclear as
the counteracting influences of donation and back-donation is under consideration.
The same reaction is observed with the alkynes as discussed for the alkenes above with a
product having an anti-isomer. An alkene is obtained as a product of the anti-addition after an
appropriate elimination of the metal. In the previous paragraph it was mentioned that
insertion into a metal hydride bond followed by hydrogenation will lead to the syn-product.
43
At this stage, it is assumed that the main characteristic and the properties that determine the
features of a catalyst are fairly recapped within the above paragraphs. We are going to
quickly discuss a metalloenzyme and biomimetic complexes which the core work mostly
depend upon as well as organometallic complexes, although we will put more emphases on
the metalloenzymes.
The organometallic catalysts usually consist of a metal centre surrounded by organic and
inorganic ligands. The properties and characteristics of an organometallic catalyst are usually
determined by both the central metal and the ligands coordination as the above discussed
characteristics clearly indicated.
1.8 Metalloenzymes
Although nature chose first row transition metals to develop metalloenzymes for biological
catalysis, second and third row transition metals are also utilised in the development of
biological systems. Metalloenzymes are protein which contains a metal cofactor that is
directly linked to the protein or to an enzyme-bound nonprotein system (Holm, R.,
Kennepohl, P., et al 1996). The utilisation of nature towards the first row transition elements
is due to the fact that the first row transition metals ions are more abundant in soluble form in
the evolution of living systems. In other terms, the metalloenzyme development is not only
predicted by optimum functionality but as well by other evolutionary pressures; for instance
by the availability of metals within the cell (Valdez C.E, Smith Q.A, et al 2014). On the other
hand, a broader collection of metals can be utilised in metalloenzyme design where the main
aim is obtaining a maximum catalytic efficiency. It is therefore essential to have a good
understanding on the binding of non-biological metals into existing metal-binding proteins
44
and its effect on enzymatic systems. Most proteins consist of metal-binding sites that contain
either amino acid side-chains or a subordinate ligand; these metal-binding sites can
accommodate non-native metals with similar coordination geometries and electronic
properties. To enhance the binding activity of the protein, the side-chain can be improved by
protein engineering while the subordinate ligands can be chemically modified. In addition to
the use of existing metal-binding scaffolds, non-native binding sites can be designed and
inserted back into the protein to create artificial metalloenzymes , as discussed in some
articles in literature (Lu Y, Yeung N, et al 2009, Petrik I.D.; Liu J, et al 2014, Yu F,
Cangelosi V.M, et al 2014, Lewis J.C, 2013).
Figure 1.12: Substitution of metal in cofactor to incorporate alternative metal into a binding
site. The Apo-protein must be synthesised first before introducing the new metal. Picture is
obtained from protein data bank 1N2C (Schindelin, H.; Kisker, C. et al 1997).
Although within these briefly discussed metalloenzymes there are a highly prominent
oxidants known as the cytochromes P450s; these are group of well-studied heme enzymes
that are known to react via oxygen atom transfer to substrates. The name P450 originated
from the fact that the strong electron donating group of the cysteinate axial ligand of the
enzyme in the carbon monoxide bound ferrous heme form gives a red-shifted robust Soret
45
band through a high-energy -* conversion of the porphyrin ring at 450 nm, hence the name
cytochrome P450. The P450s utilise a high valent iron(IV)-oxo heme cation radical species as
its active oxidant to catalyse a range of chemical reactions, such as: 1) aromatic
hydroxylation, 2) double bond epoxidation leading to an epoxide, 3) sulphoxidation reaction
mechanism and 4) aliphatic hydroxylation of a saturated compound to give either an alcohol
or an unsaturated species. The heme enzymes are subdivided into a monoxygenases and
dioxygenases (de Visser, S.P., 2009). A monoxygenase binds an oxygen molecule (O2) into
the active site of the enzyme, where it is utilised on a transition metal centre to oxidize a
substrate. In the process, a water molecule is released as well. A dioxygenase, by contrast
uses both oxygen atoms of O2, which are donated to substrate(s). These oxygen binding
enzymes are naturally classified according to their ligand system as heme or a non-heme
metal-loenzymes (Conroy-Lewis, F. M.; Mole, L.; et al 1991, Shaik, S., de Visser, S.P., et al
2004). Heme enzymes are among the most thoroughly studied enzymes in the past 50 years.
The work has gained thorough understanding of its structure, mechanisms, biological
function and its diversified isozymes. It has long been known that the P450s catalyse the
reductive activation as well as molecular oxygen scission that binds to their heme metal iron
thereby leading to the insertion of an atom of oxygen into a substrate concomitant to the
production of a water molecule from the other oxygen atom (Pan, Z., Zhang, R., et al 2006).
The reaction discussed above requires two electrons and two proton transfers to the heme
metal. Equation 1 below shows the typical P450 reaction overview of aliphatic
hydroxylation.
RH + O2 + 2e- + 2H+ R – OH + H2O (1)
46
As mentioned earlier, the cytochrome P450s catalyse a diverse range of chemical reactions
other than hydroxylation as represented in Equation 1 above; because of this there are many
P450 isozymes with different structure and folding patterns which means that all the
reactivity patterns catalysed have different proton and electron relay channels. Nevertheless,
all P450 isozymes undergo the same catalytic cycle starting from the resting state (A) as
depicted in Figure 1.13
Figure 1.13: Schematic representation of the catalytic cycle of a typical substrate
hydroxylation reaction of substrate R–H by P450 enzymes and a visual depiction of the active
species (Highlighted in a rectangular box) (Nam, W., et al., 2000).
N N
N NFe3+
O
O
S
O
ON N
N NFe3+
O
O
S
O
O
N N
N NFe2+
O
O
S
O
O
N N
N NFe2+
O
O
S
O
O
N N
N NFe3+
O
O
S
O
O
N N
N NFe3+
O
O
S
O
O
N N
N NFe4+
O
O
S
O
O
N N
N NFe3+
O
O
S
O
O
47
In the first step of the cycle, the displacement of the distal water ligand due to substrate
binding occurs to form (B). This leads to a shift in ferric heme iron spin from a low-spin (S =
½) to a high-spin (S = 5/2). The spin-change triggers an electron transfer from the redox
partner to reduce the heme iron to a ferrous state to form the iron(II) complex (C). The
ferrous heme binds oxygen to form the ferrous-oxy intermediate (D) and its reduction gives
the peroxy state (E) by the delivery of a second electron from the redox partner. Protonation
produces the ferric hydroperoxo complex known as Compound 0 (F). An extra protonation
leading to a scission of the dioxygen bound as well as the production of a water molecule is
made to form Compound I (G). The intermediate formed on the heme is a ferryl-oxo heme
cation radical species known as compound I. The compound I (structure G outlined in a
rectangular box in figure 1.12) are known to be highly reactive; the compound I and
compound 0 are also known to have a very short-lifetime and it took huge experimental
efforts to characterise and trap these species experimentally within the past few decades.
Compound I attacks a nearby substrate via hydrogen atom abstraction/radical rebound
process according to the cycle depicted above to form the alcohol product (R –OH). The
latter dissociates from the substrate binding pocket, which allows it to refill with water
molecules and a new substrate. The heme relaxes back to the resting state ferric iron thereby
completing the catalytic cycle. In some instances, the formation of non-productive pathways
leads to dead-end products; for instance the ferrous-oxy species can decay to reform ferric
P450 with production of superoxide.
In this report we will discuss work performed on the mechanism of P450 and biomimetic
model complexes. The work has given insight into the factors that determine the rate constant
of oxygen atom transfer, in particular, substrate epoxidation and has led to a predictive model
48
for substrate activation. Although we have not re-visited all the reactivity patterns of the
iron(IV)-oxo species in P450 enzymes with substrates, we have investigated a few which will
be briefly discussed and summarised within the chapter.
Figure 1.14: Reactivity patterns of P40 Compound I with selected substrates.
1.9 Epoxidation
P450 compound I is known to react via oxygen atom transfer to olefins, thereby generating
epoxides. For example, the formation of an epoxide is an initial step in the metabolism of
monosaturated fatty acids by P450 enzymes. Double bond epoxidation by P450 isozymes is
widely studied computationally (de Visser S.P, 2012, de Visser S.P., Kumar, D., et al 2004,
Shaik, S., de Visser, S.P., et al 2002). The mechanism of olefin epoxidation by P450 enzymes
49
involves a step-wise reaction via a radical intermediate (Pan, Z., Zhang, R., et al 2006, Nam,
W., 2000). The reaction starts with an electrophilic attack of the oxo group on the double
bond through a rate-determining barrier (first transition state) that leads to a C – O bond
formation between the oxo group of the iron (IV)-oxo oxidant with the carbon of the
substrate. The above mechanism gives a radical intermediate. This step is followed by a ring
closure barrier (second transition state) to form epoxide products. It is a known fact that the
ring closure barrier is negligible on the double spin state, which means the reaction is pseudo-
concerted (de Visser, S.P., 2012). The mechanism leading to the epoxidation reaction by a
substrate 2-butene is displayed in figure 1.14 using a potential energy profile to display the
energy barriers at each stage of the mechanism.
Figure 1.15: Potential energy profile for the epoxidation reaction of 2-butene by 4,2
[Fe(IV)-
oxo Porphyrin with a thiolate axial ligand as calculated with DFT methods. Data taken from
ref. (de Visser 2012).Values in Kcal/mol.
50
Epoxidation reactions of terminal olefins sometimes lead to suicidal complexes (de Visser
S.P, 2012, de Visser S.P., Kumar, D., et al 2004, Shaik, S., de Visser, S.P., et al 2002), where
a covalent bond between the heme and the substrate is formed. This ultimately leads to loss
of the metal from the heme, and hence, deactivates the enzyme. Computational modelling
showed that dead-end complexes are formed when a state crossing occurs from the radical
intermediate to a cationic state (de Visser, S.P., Ogliaro, F., et al 2001)
1.10 Hydroxylation
The substrate hydroxylation reaction is a common reaction catalysed by P450 enzymes, both
with aliphatic and aromatic substrates; although the reaction mechanisms are quite distinct. In
aliphatic hydroxylation, the reaction starts with a hydrogen atom abstraction prior to a radical
rebound to form alcohol as a product; it also proceeds via a two state reactivity pattern on
competing doublet and quartet spin states. However, in the aromatic hydroxylation the
hydrogen atom abstraction from C-H is thermodynamically difficult and requires a
substantial amount of energy to break, hence the reaction path undergoes a mechanism
known as the electrophilic substitution mechanism (de Visser, S.P., 2012). The aromatic
hydroxylation reaction studies in the past (de Visser, S.P., 2012, Bojic, M., Sedgeman, C.A.,
et al 2015) have all confirmed to have given the same pathways.
The reaction typically starts via an electrophilic attack of the oxo group on one of the carbon
atoms of the attacking aromatic substrate to form a conjoined complex known as the
“Meisenheimer complex” with a configuration of either a cationic or a radical intermediate
(de Visser, S.P. 2012). Typically compound I has a distinctive electronic configuration
consisting of three unpaired electrons: of the three unpaired electrons two are located on
51
orthogonal *FeO orbitals (*xz and *yz), whereas the third is located on the heme within an
orbital in D4h symmetrical orientation labelled a2u (Green, M. T., 1999, Ogliaro, F.; de Visser,
S. P.; 2001). The optimised geometry of compound I as depicted in figure 1.4. (de Visser, S.
P.; Shaik, S.; et al 2003, Ogliaro, F.; de Visser, S. P.; 2001), and the high-lying occupied as
well as the low-lying virtual molecular orbitals that are essential for the mechanism of
hydroxylation reaction. The a2u orbital mentioned earlier is found on the heme and it is
responsible in mixing with a lone-pair orbital on the axial ligand (Ogliaro, F.; Cohen, S.; et
al 2000).
Figure 1.16: High-lying occupied and low lying virtual orbitals of Compound I at an
optimised geometry of P450 Cpd I of quartet and doublet spin electron arrangements.
In the cationic intermediate a double electron transfer converts the active iron centre to an
oxidation state of Fe(III) with a doublet orbital occupation of x2-y22 *xz
2 *yz1 a2u
2 ФR
0 or a
quartet orbital occupation of x2-y22 *xz
1 *yz1 *z2
1 a2u
2 ФR
0,
where ФR represent the orbital
N N
N NFeIV
O
S
Cys
HOO
OHO
22 yx
2*z
xy*
yz* ua2
ua1
xz*
Quartet spin
Doublet spin
52
on the substrate . In the radical intermediate however, an electron transfer from the aromatic
substrate to the oxidant occurs hence giving a Fe (IV) type complex with an electronic
configuration of x2-y22 *xz
1 *yz1 a2u
2 ФR
1. As can be observed from Figure 1.15, the doublet
spin orbital occupation in the cationic intermediate is found to be very much lower in energy
as compared to the radical intermediate pathway. In the subsequent step the ipso-proton is
abstracted by a nitrogen atom of the porphyrin forming an intermediate of the proton transfer
which leads to the restoration of the aromaticity of the substrate ring; the proton transfer
intermediate is known to be an enormously exothermic process. In the final step of the
reaction the hydrogen atom bounded to the nitrogen atom of the porphyrin ring rebound back
to either the oxo group to form either an alcohol product or to the ortho-carbon to form
ketone.
1.11 Dissertation summary
Clearly, homogeneous catalysts have important chemical, industrial and biological functions.
However, there is only limited understanding of the molecular details of what happens during
a reaction mechanism. Evidently there are many un-answered questions on the catalytic
behaviours of verse majority of synthetic catalysts both bio-dependent and industrial catalysts
which requires detailed computational studies in order to explicitly understand there
behaviours down to an electronic level. To contribute on the understanding of these
mechanistic behaviours on a quantum level, we studied a few synthetic complexes during the
period of my doctorate degree. The quantum mechanical calculations done within this thesis
is believed to have contributed in the understanding of the mechanisms of these reactions as
53
well as the origin of the reaction pathways and environmental perturbation affecting the
mechanisms. Below is a brief description of each of the chapters elaborated within the thesis.
Chapter 3:
This chapter reported the computational results I have obtained on a synthesised crystal
structure of 1,2-bis[-hydroxo iron(III) 5-(2,3,7,8,12,13,17,18-octaethylporphyrinyl)]ethane
with I3–, ClO4
- and BF4
– counter anions which shows some unique behaviour of the system
from a spectroscopic characterisation shown by Sankar and co-workers (Ghosh, S.; Rath, S.
2010). The synthetic dimer system shows a character of two unequivalent spin states with one
iron centre having an admixed intermediate-spin state of S=3/2 with a slight contribution of
S=5/2, and a S=
5/2 on the second iron or a spin state of either S=
5/2, S=
5/2 or S=
3/2, S=
3/2 on
both of the iron centre depending on the counter anion approaching. The spectroscopic
characterisation reported that, as the approaching counter anion altered, the spin state
ordering on the -hydroxo complexes changes. In order to fully understand and rationalise
how the spin state ordering is affected by external perturbations, we also have done a
comprehensive computational benchmarking which was also thoroughly discussed in the
chapter. Our calculations demonstrated that subtle environmental perturbations, such as
entropic corrections ruffles the spin state ordering and relative energies and are likely to be
the root cause of the variation in spin state ordering observed experimentally.
Chapter 4:
In this chapter, I have synthesized a model complex and studied it with low-pressure Fourier
transform-ion cyclotron resonance (FT-ICR) mass spectrometry (MS). I found out that all
54
substrates react with the selected substrates by a more or less efficient oxygen atom transfer
process; Whereas substrates with low ionization energies react by a charge-transfer channel,
which enabled me to determine the electron affinity of [FeIV(O)(TPFPP+•)]+ for the first
time. The computational results we presented here confirm the observed trends excellently
and rationalize the reactivities within the framework of thermochemical considerations and
valence bond schemes.
In summary my findings gave an excellent agreement with ideal-gas measured rate constants
and have provided a general trend in epoxidation reactions.
Chapter 5:
This chapter basically report a systematic study into the core chemical properties of the
Cytochrome P450 oxidant and substrates in order to understand the factors that affect
reactivity patterns. The work purposely focused on investigating the effect bestowed by an
epoxidation of styrene and a para-substituted styrene substrate by a biomimetic compound I
with either an anionic or a neutral axial ligand. We chose chloride as our anionic ligand and
an acetonitrile for our cationic axial ligand. Our findings reported that the activation enthalpy
of the reaction is determined by the ionization potential of the susbstrate, the electron affinity
of the oxidant and the C-O bond strength which is determined by the bond dissociation
energy (BDEOH). We also generated a model that enable us to predict the rate constants and
reactivities of substrates epoxidation reactions by iron(IV)-oxo porphyrin cation radical
oxidants.
In summary the discussion presented within the chapter states that electron withdrawing
substituents lead to early transition states while electron donating groups on the substrates
55
offers late transition states. This trends affects the barrier heights in such a way that electron
withdrawing substituents correlate the barrier height with BDEOH, while the electron affinity
of the oxidant is proportional to the barrier height for substrates with electron donating
substituents.
Chapter 6:
The calculations within this chapter gave a multistate reactivity patterns, whereby the
bifurcation of hydroxylation versus desaturation was detailed. The product distributions differ
on each of the spin state surfaces; hence we find spin-selective product formation. I did a
thorough analysis of the electronic and thermochemical factors that determine the bifurcation
pathways and I come up with a model that predicts the regioselectivity of the aliphatic
hydroxylation over desaturation pathways from valence bond and molecular orbital theories.
Chapter 7:
This chapter summarises the work I have done in order to establish the key features of a
synthetic homogeneous catalyst that is capable of catalysing an alkyl chain growth. I have
shown some detailed results we achieved on a combined DFT, an NBO and a VB analysis;
and also a detailed thermochemical studies has been performed and establish the intrinsic
properties of the complex. Initially I analyse the reactants as having a Ru2Pt-carbene core
where the carbene’s lone-pair transfers to the Pt through bonding configuration. The
characterisation also shows that a low energy mechanism of alkyl formation via consecutive
CH3• followed by H• transfer to the bridging carbene to form alkyl chain is feasible.
Conclusively, my work identifies a novel catalyst for the synthesis of alkanes that starts from
56
a trimetal carbene. This unique structure is shown to be capable of intramolecular methyl and
hydrogen atom transfer to the carbene to form 2-CHCH3 products as precursor to alkanes.
Chapter 8:
The computational analysis presented in this chapter gave an extended knowledge on the
importance of distal hydrogen bonding on both heme and non-heme complexes. I show for
the first time, the effect of hydrogen bonding interaction with the distal Oxo ligands. It is
known that Iron (IV)-oxo intermediates are actively involved in several hydrogen bonding
interactions with its neighbouring molecules in an active site of an enzyme which triggers our
curiosity to gain a better understanding on it. The calculation therefore gave a new insight
into the distal hydrogen bonding and it shows that the hydrogen bond may play a vital role in
proton relay mechanisms in the formation of metal-oxo intermediates thereby decreasing the
hydrogen atom abstraction ability of the intermediate. Indeed, in nonheme iron enzymes,
where no proton relay takes place, there generally is no donating hydrogen bond to the
iron(IV)-oxo moiety.
57
CHAPTER 2
METHODOLOGY
58
Computational techniques have been developed with the aim to predict and improve
understanding of molecular behavior. The widely known and accepted procedures used in the
calculation of atomistic systems include ab-initio methods, in which a system is calculated
from scratch without any prior knowledge of the chemical system, whereas in
(semi)empirical Quantum mechanics methods experimentally derived parameters are taken
into account and build upon. By contrast, Molecular mechanics method (MM) are commonly
used for calculations of very large systems (typically with tens of thousands of atoms, such as
proteins). However, the MM methods do not model bond-formation and bond breaking
processes. In all methods aforementioned, the accuracy of the result solidly depends on the
computational theory applied. In order to combine the accuracy of QM methods with the
speed of MM methods, the combined quantum mechanics/molecular mechanics (QM/MM)
method has been developed. QM methods generally are computed based on the Schrödinger
equation (equation 2.0), which was introduced by the Austrian physicist Erwin Schrödinger
(Schrödinger, E., 1926), in 1926 and is the basic foundation of the electronic structure
calculations.
H = E
In Eq 2.0, H denotes the Hamilton operator, represents the eigenfunction for a given
Hamiltonian (i.e. the wavefunction), and E is the energy of the system. The wavefunction
takes the position of the electrons and nuclei in the system as variables, leading to the
following equation:
𝐻 Ψ 𝑖 (𝑥1, … , 𝑥𝑁 , 𝑅1, … , 𝑅𝑀) = 𝐸 Ψ𝑖(𝑥1, … , 𝑥𝑁 , 𝑅1, … , 𝑅𝑀) (2.1)
59
The position of the electrons in Eq 2.1 is described with 𝑥𝑁, while N and 𝑅𝑀 describe the
position of the nuclei, M. The properties of the system will be assumed using the knowledge
of 𝛹 in the equation.
The wavefunction is assumed to be orthonormal in space as represented in the equation 2.2
below (Schrödinger, E., 1926):
⟨Ψ𝑖|Ψ𝑗⟩ = 𝛿𝑖𝑗 (2.2)
The term 𝛿𝑖𝑗 is known as the Kronecker symbol, with a range as indicated 𝛿𝑖𝑗 = 1 𝑖𝑓 𝑖 =
𝑗 and 𝛿𝑖𝑗 = 0 𝑓𝑜𝑟 𝑖 ≠ 𝑗.
2.1 Solving the Schrodinger equation
To solve the Schrodinger equation, the Hamilton operator needs to be established in the
following form (Schrödinger, E., 1926):
𝐻 = 𝑇𝑒 + 𝑇𝑛 + 𝑉𝑛𝑒 + 𝑉𝑒𝑒 + 𝑉𝑛𝑛 (2.3)
Where the terms 𝑇𝑒 and 𝑇𝑛 representing the terms for kinetic energy of the electrons and
nuclei, The value 𝑉𝑛𝑒 is the attractive potential between the nuclei and the electrons, and the
potentials 𝑉𝑒𝑒 and 𝑉𝑛𝑛 represent the electron-electron and nuclei-nuclei repulsion energies
respectively. The individual terms can be equated into the following expressions:
𝑇𝑒 = −1
2 ∑ ∇𝑖
2
𝑁
𝑖=1
(2.4)
𝑇𝑛 = −1
2 ∑
1
𝑀𝐴∇𝐴
2
𝑀
=1
(2.5)
60
𝑉𝑛𝑒 = − ∑
𝑁
𝑖=1
∑𝑍𝐴
𝑟𝑖𝐴
𝑀
𝐴=1
(2.6)
𝑉𝑒𝑒 = − ∑
𝑁
𝑖=1
∑1
𝑟𝑖𝑗
𝑁
𝑗>1
(2.7)
𝑉𝑛𝑛 = − ∑
𝑀
𝐴=1
∑𝑍𝐴𝑍𝐵
𝑅𝐴𝐵
𝑀
𝐵>𝐴
(2.8)
Substituting the above terms into equation 2.3 above gives the following equation 2.9:
𝐻 = −1
2 ∑ ∇𝑖
2
𝑁
𝑖=1
−1
2 ∑
1
𝑀𝐴∇𝐴
2
𝑀
𝐴=1
− ∑
𝑁
𝑖=1
∑𝑍𝐴
𝑟𝑖𝐴
𝑀
𝐴=1
+ ∑
𝑁
𝑖=1
∑1
𝑟𝑖𝑗
𝑁
𝑗>1
∑
𝑀
𝐴=1
∑𝑍𝐴𝑍𝐵
𝑅𝐴𝐵
𝑀
𝐵>𝐴
(2.9)
The A and B denote the two individual atoms in molecule M (or AB), whereas i and j
represent the N number of electrons within the system.
The equation above is the full Hamiltonian for a bimolecular system where the first
expression represents the kinetic energy contribution of the electrons; the second expression
represents the kinetic energy of nuclei M, the third expression is the electron-nucleus
attraction, the fourth expression is the nucleus-nucleus association and the last is representing
the electron-electron repulsion (Schrödinger, E., 1926). However, as mentioned earlier, the
expression in equation 2.9 above gets more complex according to the size of the system. This
equation is the basis of the electronic Schrödinger equation and is solved in Quantum-
chemical calculations.
61
2.2 Born-Oppenheimer approximation
The Born-Oppenheimer approximation is the foremost of quite a few approximations used in
quantum chemistry to simplify solving the Schrödinger equation. The molecular challenge
encountered by the Schrödinger equation is simplified by separating nuclear and electronic
motion by this approximation which is reasonable due to the fact that the mass of the electron
is 1822 times smaller than the mass of the proton (Millikan R, A., 1911). Therefore the nuclei
move very slowly with respect to the movement of the electrons, and the electrons react
immediately to changes in nuclear position. Consequently, the electron distribution within a
molecular system will depend on the positions of the nuclei, and not on their velocities.
Literally, the nuclei are fixed with respect to the position of the electrons, and electronic
motion can be described as occurring in a field of fixed nuclei. This gives rise to a
wavefunction of an electron that describes the electrons in the field of the nuclei; therefore
the Hamiltonian is defined as the following equation (Atkins, P., W. and Friedman, R., S.
1997):
𝐻𝐵𝑂 = −1
2 ∑ ∇𝑖
2 − ∑ ∑𝑍𝐴
𝑟𝑖𝐴+ ∑ ∑
1
𝑟𝑖𝑗 + 𝑉𝑛𝑚
𝑁
𝑗>1
(2.10)
𝑁
𝑖=1
𝑀
𝐴=1
𝑁
𝑖=1
𝑁
𝑖=1
The 𝑉𝑛𝑚 is a constant representing the nucleus-nucleus repulsion; as can be observed the
equation 2.10 can factorized to give equation 2.11 below;
𝐻𝑒𝑙𝑒𝑐 = ∑ (−1
2 ∇𝑖
2 − ∑𝑍𝐴
𝑟𝑖𝐴+ ∑
1
𝑟𝑖𝑗
𝑗>𝑖𝐴
) (2.11)
𝑖
62
Now, the wavefunction depends solely on the number of electrons in the system as indicated
below:
𝐻𝑒𝑙𝑒𝑐 Ψ 𝑖(𝑒𝑙𝑒𝑐) (𝑥1, 𝑥2, … , 𝑥𝑖, 𝑥𝑗 , … , 𝑥𝑁) = 𝐻𝑒𝑙𝑒𝑐 Ψ 𝑖(𝑒𝑙𝑒𝑐)(𝑥1, 𝑥2, … , 𝑥𝑖, 𝑥𝑗 , … , 𝑥𝑁) (2.12)
The electrons can be described by their spin quantum number (ms) which can take up a value
of either – ½ or ½; the definition is based upon the alignment of the spin with respect to an
arbitrary axis; where x represent the probability of finding an electron at a given point in
space. The aforementioned spins are known as and they are spinfunctions for ms = + ½
and -½ correspondingly; and are considered orthonormalised:
⟨𝛼|𝛼⟩ = ⟨𝛽|𝛽⟩ = 1 (2.12)
⟨𝛼|𝛽⟩ = ⟨𝛽|𝛼⟩ = 0 (2.13)
Generally, the wavefunction is split in to one-electron wavefunctions, that contain a spatial
component and a spin component as represented in equation 2.14:
Ψ (𝑥) = Ψ (𝑟). 𝜎 𝜎 = 𝛼 𝑜𝑟 𝛽 (2.14)
63
2.3 Slater determinant
2.3.1 Anti-symmetry Principle
The wavefunction Ψ is not visible; however it is expressed according to the representation in
equation 2.15 (Atkins, P., W. 1977);
|Ψ(𝑥1, 𝑥2, … , 𝑥𝑁)|2𝑑𝑥1𝑑𝑥2 … 𝑑𝑥𝑁 (2.15)
Where 𝑑𝑥1𝑑𝑥2 … represent the probability of finding an electron at a specific point in space.
As electrons are equivocal and indistinguishable, the probability does not change by the
exchange of two electrons;
|Ψ(𝑥1, 𝑥2, … , 𝑥𝑖𝑥𝑗 , … , 𝑥𝑁)|2
= |Ψ(𝑥1, 𝑥2, … , 𝑥𝑖𝑥𝑗 , … , 𝑥𝑁)|2
(2.16)
However, the sign of the wavefunction is changed because of the exchange of two electrons;
the Ψ becomes anti-symmetrical with respect to the electron changes which are a clear
representation of the quantum-mechanical generalization of the Pauli’s exclusion principle
which states that no two electrons can occupy the same state within a chemical system (Krane
K. S., 1987). Since the precise wavefunction is unknown, a trial wavefunction that obeys the
anti-symmetrical principle is carefully generated. A trial wavefunction is generated and
conforms to the rule with anti-symmetry product of N one-electron wavefunctions 𝑥𝑖(𝑥𝑖).
This product is represented by the symbol Φ𝑆𝐷 and is referred to as the Slater determinant
(Atkins P. W., 1977):
Φ𝑆𝐷 = 1
√𝑁! |
𝑥1(𝑥1) 𝑥2(𝑥1) … 𝑥𝑁(𝑥1)
𝑥1(𝑥2) 𝑥2(𝑥2) … 𝑥𝑁(𝑥2)
𝑥1(𝑥𝑁) 𝑥2(𝑥𝑁) … 𝑥𝑁(𝑥𝑁)| (2.17)
64
The columns in the representation above are single electron wavefunctions (orbital), x(x),
whereas the row represents the electron indices.
In order to establish the best wavefunction to describe a chemical system, the variational
method (also known as Rayleigh-Ritz variational method) is used. It is based on the two most
pronounced methods known as the Hartree-Fock method and Density functional theory
(DFT) (Abdulsattar, Mudar A. 2012). The variational method states that the value calculated
for the total energy of a trial wavefunction can only be greater than or equal to the ground
state energy 𝐸0.
𝐸 = < Ψ|𝐻|Ψ > ≥ 𝐸0 (2.18)
2.4 Hartree-Fock theory (HF)
In Hartree-Fock theory, the Hamiltonian is typically by two individual components: 1) The
core Hamiltonian, Hc(i), which describes the kinetic energy of the electrons as well as the
electron-nucleus attraction, and 2) The electron-electron repulsion:
𝐻 = ∑ [𝐻𝑐 (𝑖) + ∑1
𝑟𝑖𝑗𝑗>𝑖
]
𝑖
𝑎𝑛𝑑 𝐻𝑐(𝑖) = −1
2 ∇𝑖
2 − ∑𝑍𝐴
𝑟𝑖𝐴 (2.19)
𝐴
The core Hamiltonian can be solved exactly as it is, but the electron-electron repulsion must
be treated in such a way that each electron is considered to be moving independent of other
(neighboring) electrons in an average field created by the other electrons. By implementing
the variational method to a single Slater determinant Φ𝑆𝐷, the lowest energy calculation is
possible through the optimsation of the values for 𝑥𝑖. The equation obtained is called
65
“Hartree-Fock equation (HF)” and solving the HF equation determines the best spin orbitals
at minimized energy E.
𝑓𝑖𝑥𝑖 = 𝜖𝑖𝑥𝑖 (2.20)
In Eq 2.20, 𝑥𝑖 is an eigenfunction of a Fock operator 𝑓, and 𝜖𝑖 is the corresponding orbital
energy for this eigenstate. The eigenvalue of the Fock operator that is negative, i.e. associated
with a spin orbital,−𝜖𝑖, actually corresponds to the ionization potential according to
Koopmans Theorem (Koopman, Tjalling, 1934). The Fock operator is an effective single-
electron operator with the form as follows:
𝑓𝑖 = −1
2 ∇𝑖
2 − ∑𝑍𝐴
𝑟𝑖𝐴+ 𝑉𝐻𝐹(𝑖)
𝐴
(2.21)
In Eq 2.21, the term 𝑉𝐻𝐹(𝑖) is known as the Hartree-Fock potential and it is a representation
of the average repulsive potential by each electron due to the field of N-1 electron. It replaces
the 1
𝑟𝑖𝑗 repulsion operator which was considered too complex to solve. Although by adjusting
the electronic repulsion of different electrons within an average potential, the equation is
solvable. Therefore 𝑉𝐻𝐹comprises of two terms:
𝑉𝐻𝐹(𝑥1) = ∑ (𝐽𝑗(𝑥1) − 𝐾𝑗(𝑥1))
𝑗
(2.22)
𝐽𝑗(𝑥1) = ∫|𝑋𝑗(𝑥2)|2
1
𝑟12 𝑑𝑥2 (2.23)
66
𝐾𝑗(𝑥1)𝑋𝑖(𝑥1) = ∫ 𝑋𝑖 (𝑥1)1
𝑟12 𝑋𝑖 (𝑥2)𝑑𝑥2 𝑋𝑗 (𝑥1) (2.24)
The operator 𝐽𝑗 represents the potential experienced by an electron at a point 𝑥1 because of
the average charge distribution of another electron within the spin orbital 𝑋𝑗, while the term
𝐾𝑗 has no known classical interpretation. It leads to an exchange of the variables within the
two spin orbitals. To solve the Hartree-Fock equation, the two aforementioned methods can
be applied depending on the instance faced. If a system is closed shell, i.e. containing an even
number of electrons, all the electrons are paired up and the Restricted Hartree-Fock (RHF)
method is used to solve it. However, if there is an odd number of electrons or an even number
of electrons with an unpaired electrons arrangement within the orbitals, the Unrestricted
Hartree-Fock (UHF) can be used. The Roothaan-Hall equation is generally used to solve the
RHF equations, whereas in the case of the UHF, the Berthier-Pople-Nesbert equation is used
to solve the integrals. To solve the 𝑉𝐻𝐹 description of equation 2.22 above, it is essential to
obtain 𝑋, therefore an iterative methodology known as the “Self-Consistent Field method
(SCF)” is required. An initial guess for 𝑉𝐻𝐹 is required to calculated 𝑋 which is used to
generate a new 𝑉𝐻𝐹. The process continues until the cycle is converged based on the criteria;
i.e. until successive potentials are identical.
2.5 Electron Correlation
In the Hartree-Fock (HF) method, the electrons are considered to be moving freely within an
average electronic field of the nuclei so the correlated motion of individual electrons
perturbed by the other electrons is omitted. As a consequence, the total HF energy is always
higher than the real energy (Abdulsattar, Mudar A. 2012). The difference between the
67
Hartree-Fock energy and the real energy is the correlation energy and the energy gap is the
electronic correlation energy.
𝐸𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛 = 𝐸𝑡𝑜𝑡𝑎𝑙 − 𝐸𝐻𝐹 (2.25)
The electron correlation term constitutes only a tiny proportion of the overall energy of the
system, however, it has a large impact on the calculated chemical and physical properties of
the system (Riley, K. E.; Op’t Holt, B. T.; et al 2007). The correlation energy is usually
introduced by taking into account the excitation of one or more electrons from occupied
orbitals to virtual orbitals. The electrons then adapt a specific orbital arrangement that
depends on the number of electrons involved within the system and the Slater determinant is
used to describe each of these states. The sum combination of all states gives the new trial
wavefunction which is considered closer to the real system than the initial determinant.
Ψ = 𝑐0 Φ𝐻𝐹 + ∑ 𝑐𝑖Φ𝑖 (2.26)
𝑖=1
In Eq 2.26, the summation is over all probable excited states. 𝑐𝑖 are the coefficients that
define the contribution of each excited state to the wavefunction.
2.6 Density Functional Theory (DFT)
Density functional theory (DFT) is based on electron densities instead of wavefunction. An
essential step on the way to developed was taken by Slater in the 1950s, where he applied his
XR method (Extrapolated Run method) for molecules and solid state systems. Thus, one-
parameter approximate exchange correlation functional (Slater, Morton. 1950) was used.
68
However, DFT did not become a full-fledged theory until the formulation of the Hohenberg
and Kohn theorems in 1964 (Hohenberg, P.; Kohn, W. 1964). They introduced orbitals into
the picture the Kohn-Sham formalism, which gave a computational breakthrough which was
later implemented in e.g., Pople’s GAUSSIAN software packages (Frisch, J. A. et al 1986).
Although, there are a number of popular and highly accuarate wave function packages in use,
DFT was further developed as a computationally attractive alternative.
Density functional theory provides a relative efficient and unbiased tool with which to
compute the ground state energy of realistic models of mass materials and their surfaces. The
calculation solidly relies on the development of some approximations for the exchange-
correlation energy functional (Harrison, J.G., 1987). However, remarkable advances have
been achieved in the recent decade in the development of an exchange-correlation functional
that is dependent on local density gradients, as well as semi-local measures of the density and
nonlocal exchange functional (Frisch, J. A. 2009). Density functional theory describes the
energy of a system as a function of the electron density which is in contrast to wave function
quantum mechanics (Frisch, J. A. 2009). Density functional theory method is widely accepted
and proven as a very successful and computationally less expensive as compared to other
methods, it also gives a result that reasonably agree with experiment for relatively large
chemical systems. However, it is less accurate for binding energies and details of the energy
of the surface away from equilibrium geometries. The total energy of a DFT calculation of
the electronic energy 𝐸𝑒𝑙𝑒 and is expressed by equation 2.27 below (Parr, R. G.; Yang, W.
1989);
𝐸𝑒𝑙𝑒 = −1
2∑ ∫ ∅𝑖(𝑟1)
𝑖
∇2∅𝑖(𝑟1)𝑑𝑟1 + ∑ ∫𝑍𝐴
|𝑅𝐴 − 𝑟1|𝐴
𝜌(𝑟1)𝑑𝑟1 +1
2∫
𝜌(𝑟1)𝜌(𝑟2)
|𝑟1 − 𝑟2|𝑑𝑟1𝑑𝑟2 + 𝐸𝑥𝑐 (2.27)
69
The first term expresses the kinetic energy of the system, the second term represents the
attraction between electron and nucleus in the system, the next is the Coulomb interaction
between two individual electrons, whereas the last term represents the exchange-correlation
functional (Exc) (Ziegler, 1991). In Eq 2.27, 𝜌(𝑟) is the electron density, r is a term
representing the interaction between electron-electron, electron-nucleus. Thus, in DFT the
solution to the electronic energy (𝐸𝑒𝑙𝑒) follows from the sum of the integrals given in equation
2.27. In addition to the kinetic energies of the electrons, Coulumbic electron-nucleus
attraction, and electron-electron repulsions, there is an exchange-correlation term 𝐸𝑥𝑐 for the
electron pair interaction with an individual exchange and a correlation component Ex and Ec
respectively.
The exchange-correlation functional is unknown, therefore, it is approximated to determine
its contribution to DFT calculations. The exchange-correlation component is divided into an
exchange energy term (equation 2.28) also known as the Slater exchange, and the correlation
energy usually taken from Vosco, Wilk and Nusair (equation 2.29).
𝐸𝑥𝑆𝑙𝑎𝑡𝑒𝑟 = −
9
4𝛼𝑒𝑥(
3
4𝜋)
13
∑ ∫[𝜌1𝛾(𝑟1)]
43
𝛾
𝑑𝑟1 (2.28)
𝐸𝑐𝑉𝑊𝑁 = ∫ 𝜌
1 (𝑟1)휀𝑐[𝜌
1𝛼 (𝑟1), 𝜌
1𝛽 (𝑟1)]𝑑𝑟1 (2.29)
The exchange scale factor αex in the equation has the value 2/3 for an electron gas (Parr, R. G.;
Yang, W. 1989) and the correlation is calculated per electron in a gas from the correlation
70
functional 휀𝑐 |𝜌1𝛼 , 𝜌1
𝛽| using spin densities 𝜌1
𝛼 and 𝜌1𝛽
. The main reason behind the use of the
Slater exchange 𝐸𝑥𝑆𝑙𝑎𝑡𝑒𝑟and the correlation energy 𝐸𝑐
𝑉𝑊𝑁 is to obtain a wavefunction-type
method, although it was found not to be good enough to compete with other methods such as
MP2; therefore, further corrections due to non-local interactions were required. These were
implementation by, for instance, Lee, Yang and Parr (LYP) and Perdew and Wang (PW95) for
correlation methods and Becke for exchange methods (Becke, A. D. 1996, Lee, C.; Yang, W.
et al 1988).
2.7 Functionals
2.7.1 Local Density Approximation (LDA)
The exchange-correlation functionals are based upon the local density approximation (LDA),
which use the electron density of a uniform electron gas. Even though the LDA method is a
rough approximation, it is the one system that is known to define the density as 𝜌 = 𝑁
𝑉
(where the number of electrons is represented as N, and volume of the gas is denoted as V ),
and the exchange as well as the correlation energy functional are known to a very high
accuracy. The electronic density is 𝜌 is replaced by the spin electronic densities 𝜌𝛼 and 𝜌𝛽
such that 𝜌 = 𝜌𝛼 + 𝜌𝛽 when dealing with an open shell systems. This approximation is
termed the Local Spin-Density Approximation (LSDA). One known instance of an LDA type
functional is the one generated by Vosko and co-workers based on high quantum Monte Carlo
calculation for a uniform electron gas. More accurate results than those obtained from HF
theory are calculated when molecular properties of a system such as vibrational frequencies,
charge moments, elastic moduli etc, are determined. However, generally, the LDA functional
71
gives occasionally serious flaws and poorly characterizes energetic details, including bond
energies and reaction mechanisms.
2.7.2 Generalized Gradient Approximation (GGA)
Although the LDA method discussed above is known to calculate the exchange and
correlation energies to a very high accuracy, it actually describes energies rather badly, and
hence a new type of functional was introduced known as the “Generalised Gradient
Approximation (GGA)”. This functional uses the gradient of the electron density ∇𝜌. The
inclusion of the electron density gradient assist in describing the non-homogeneity of electron
density rather more realistically. The GGA is usually sectioned into an exchange term and a
correlation term, which can then be solved independently:
𝐸𝑋𝐶𝐺𝐺𝐴 = 𝐸𝑋
𝐺𝐺𝐴 + 𝐸𝐶𝐺𝐺𝐴 (2.30)
2.7.3 Hybrid Functionals
Computational chemistry was greatly enhanced by introduction of the hybrid density
functional procedure developed by Becke. He compared information from experimental test
sets, namely ionization energies, electron and proton affinities, and compared it against DFT
calculated values. The result gave a value with high accuracy which demonstrated the
accuracy of the Hybrid functional method. Although, many different combinations between
exchange and correlation functionals are possible, currently the most commonly used one is
the Becke 3-Parameter (Exchange) Lee, Yang and Parr (correlation; density functional
theory) B3LYP (Lee, C.; Yang, W. 1988). The hybrid density functional method B3LYP is
expressed as the following equation:
72
EXCB3LYP = AEX
Slater + (1 + A)EXHF + BΔEX
Becke + ECVWN + CΔEC
LYP (2.31)
Hence, the equation comprises of the Slater and Vosko-Wilk-Nusair local density
approximation functions, the Hartree-Fock exchange, and a correction term for the exchange
due to Becke and Lee, Yang and Parr’s correlation functional (Becke, 1993). As mentioned
earlier the B3LYP hybrid functional is a mixture of the LDA and GGA functional, as
presented in equation 2.32 below:
EXCB3LYP = EXC
LDA + 𝑎0 (EXHF − EX
LDA) + 𝑎𝑋 (EXGGA − EX
LDA) + 𝑎𝐶 (𝐸𝐶𝐺𝐺𝐴 − 𝐸𝐶
𝐿𝐷𝐴) (2.32)
Where the terms a0 = 0.20, 𝑎𝑋 = 0.72 and 𝑎𝐶 = 0.81 are three empirical parameters
determined by fitting the predicted values to a set of atomization energies, proton affinities,
ionization potentials and total atomic energies of a system (Becke, 1993).
2.8 Basis Sets
Atomic orbitals and molecular orbitals are described and created using basis functions. They
are denoted as a linear combination of such functions with the coefficients to be determined.
The basis-functions are assumed to be obtained at the atomic nuclei at the centre of the atom
and so bear some resemblance to an atomic orbital. The basis functions are categorized into a
Slater-type orbital (STO), and the Gaussian-type orbital (GTO).
2.8.1 Slater-type Orbital (STO)
The Slater-type orbitals (STO) have the exponential dependence 𝑒−𝜁𝑟 and are very close in
their mathematical expression to the real atomic orbital. It is also worth mentioning that
solving the Schrodinger equation for a H-atom gives the Slater-type orbitals.
73
휂𝑆𝑇𝑂 = 𝑁𝑟𝑛−1𝑒−𝜁𝑟𝑌𝑙𝑚(휃, 𝜙) (2.33)
Where the term N represents a factor of normalization, 휁 is the exponent; 𝑌𝑙𝑚is the angular
momentum which is the function describing the shape of the orbital; r, 휃 and 𝜙 are spherical
coordinates, while n, l and m are classical quantum numbers representing principal, angular
and magnetic momentum respectively.
The STO very closely described the behavior of atomic orbitals of hydrogen because it
features a cusp at r = 0 and a good exponential decay for larger values of r.
2.8.2 Gaussian-type orbital (GTO)
The Gaussian-type orbitals (GTO) have the exponential dependence𝑒−𝛼𝑟2:
휂𝐺𝑇𝑂 = 𝑁 𝑥𝑙𝑦𝑚 𝑧𝑛 𝑒−𝑎𝑟2 (2.34)
Where N is a normalization factor, x, y and z are the Cartesian coordinates. Although
Gaussian-type orbitals are not really orbitals as they are simpler functions and are frequently
referred to as the “Gaussian Primitive” (Lowe, John P. 1978). Gaussian primitives are
usually obtained from quantum calculations on Hartree-Fock or Hartree-Fock plus some
correlated calculations. Typically the exponents of x, y and z are varied until the lowest total
energy of the atom is attained. For molecular calculations, these Gaussian primitives have to
be contracted thereby the Cartesian linear combinations of them will be used as basis
functions. The basis functions will have its coefficients and exponents fixed.
The GTO, in contrast to the above STO explained, does not show a cusp at r = 0 and steeply
decrease for large values of r. Despite the challenges mentioned, the GTO are better because
the product of two GTOs centered on two different atoms is a third GTO sandwiched between
74
them, which is not the case with an STO basis function and that made them very difficult to
handle computationally as the four-centre-two-electron integrals are time consuming.
The precision and degree of complexity of a basis set is defined by the number of contracted
functions involved in representing each atomic orbital, the minimum being one contracted
function to describe a basis function. For instance, the basis set STO-3G is formed by a linear
combination of three contracted Gaussian function (represented by G) so as to resemble one
STO. At least, usually two or more basis functions can be used to describe each type of
orbital for more precision and better descriptions. The valence electrons are known to be
actively involved in chemical reactions and therefore, it is essential to have a flexible
description of its electrons. Such basis sets with valence electrons is defined with more
flexibly and differently from the core orbitals and are generally referred to as “Split Valence
Basis Sets”. An example of a split valence basis sets most commonly used in electronic
structure calculations is the 6-31G basis set. These type of basis sets are represented in the
format X-YZG, with nomenclature as follows:
X is the number of primitive GTOs used in describing one single contracted Gaussian
function of the core.
Y and Z are the number of primitive GTOs describing the valence orbitals; more can
be added to increase precision.
Therefore, considering the above nomenclature, the split valence basis set mentioned earlier
as 6-31G comprises two functions, one containing three primitive and the other contain only
one.
75
2.8.3 Polarisation and Diffuse basis functions
Polarization and diffuse functions can also be included to a basis sets. The polarization
represents the deformation of the electronic cloud, which is induced by the bond between two
atoms. To define this characteristic, functions with higher angular momentum are added to
the basis set. For instance, the addition of a p function to a hydrogen atom induces
polarization; the same applies when a d function is added to a basis set containing p valence
orbitals, and an f function for a d-valence orbitals. Even more precision is increased by the
addition of for instance, p and d polarization functions to a hydrogen atom with 6-31G basis
set thereby becoming a 6-31G(pd) basis set.
The diffuse function, on the other hand explains the section of atomic orbitals distant from
the nuclei that could play a very important role when considering anions or diffuse electronic
clouds; for instance, in second or third row transition metals. The diffuse function was also
found useful when included in calculating anions containing lone pairs.
Another circumstance worth mentioning is the Effective core potential used to describe the
inner core of transition metal. Transition metals have a very large inner core and so the
number of basis functions used to describe it would be very large and that could hinder with
the accuracy of the calculations hence the application of the effective core potential comes in
handy. The ECP will model the effect of the nucleus and the electrons from the inner shell on
the valence electrons as an average effect; these reduce the computational time and includes
the relativistic effects on the system studied as these basis functions are generated from
relativistic atomic calculations.
76
2.9 Zero-Point Energy
The Born-Oppenheimer approximation is almost always used in the field of quantum
chemistry (Lowe, John P. 1978). Therefore, the most common definition of the molecular
zero-point energy (ZPE), the energy difference between the vibrational ground state and the
lowest point on the Born-Oppenheimer potential energy surface (Lowe, John P. 1978).
However, the definition is not applicable to experimental ZPE as the Born-Oppenheimer
approximation (BOA) is never taken on by real molecules. Experimentally, spectra are
analyzed, though the explicit Born-Oppenheimer approximation correction is only made
when simultaneously fitting of different isotopologs to effective potential. Therefore, the
experimental zero-point energy is defined as the difference between the molecular ground
state and the lowest point on its isotope-specific effective potential. The variation between the
aforementioned definition and the definition used by the quantum chemist must be absorbed
by the empirical scaling factor that is applied in order to determine zero-point energy
theoretically (K.Irikura, 2006).
The computational procedure of determining the ZPE is by solving the secular determinant
which gives an orbital energies that sum up the electronic energy (Ee), that represent the
ground state energy of the gaseous system at 0K temperature. However, since the nuclei
occupy vibrational energy levels, the energy obtained is not the correct absolute energy,
therefore, the electronic energy needs to be corrected for ZPE. Computationally, the ZPE
correction is found from a geometry optimization followed by a vibrational frequency
calculation which estimates the vibrational energy levels and susbsequntly the ZPE value.
The energy of the lowest vibrational energy level is expressed with equation 2.35 below:
77
𝐸0 = 𝐸𝑒 + 𝑍𝑃𝐸 (2.35)
In Eq 2.35, 𝐸0 is the actual (ZPE corrected) energy of the system at 0 K and 𝐸𝑒 is the
electronic energy which represents the ground state energy of the system at 0 K temperatures.
A frequency calculation can also be used to determine the nature of a stationary point found
by a geometry optimization. There are two portion of information from the frequency output
which are essential to typifying a stationary point (Foresman, J., B. 1993):
The number of imaginary frequencies
The number of negative eigenvalues in the Hessian
Imaginary frequencies are listed in the output of a frequency calculation as negative numbers.
A structure with N number of imaginary frequencies is an nth
order saddle point (Foresman,
J., B. 1998). When a structure contains an imaginary frequency, the calculation indicates that
there is some geometric distortion for which a lower energy structure can be found.
Sometimes, a calculation stuck in a high symmetry constraint and lowering the symmetry of
the structure leads to a more stable structure (Foresman, J., B. 1993). To differentiate between
these two cases, the eigenvalue of the Hessian is taken under consideration.
2.10 Transition states
A transition state refers to the lowest energy maximum state, which connects to two local
minima; it should be a first order saddle point with one maximum along one degrees of
freedom (i.e. the reaction coordinate), and a minimum energy along all other degree of
freedom. Experimentally, transition states are not visible and cannot be characterized as they
have a very short lifetime. However, the activation energy (the energy between the transition
state and the reactant) connects to the rate constant for the chemical reaction. Transition state
78
calculations are done basically by first running a geometry scan between the reactant and
product geometries, subsequently, the geometry scan with the highest energy is taken and a
full transition state search is done; following with a frequency calculation will characterize
the transition state as a first order saddle point with one imaginary frequency for the correct
mode. It is done by fixing specific intervals while at the same time the rest of the structure
will be fully optimized
Figure 2.1: A Potential Energy Surface (PES) indicating a reaction from reactant to product
that passes through a transition state (TS).
2.11 Computational Software
All calculations implemented within this thesis is performed using electronic structure
method software packages such as Gaussian03 (Frisch, M. et al 2003), Gaussian09 (Frisch,
M. et al 2009), Jaguar 7.6 (Schrödinger LLC, 2007), Orca (Neese, F, 2009), Turbomole
(Turbomole V6.2 2010) and the Chemishell Interface (Sherwood, P. et al 2003) series of
Programs and used without modifications.
79
CHAPTER 3
PROJECT ONE
80
Spin state ordering in -hydroxo bridged
diiron(III)bisporphyrin complexes1
ABSTRACT
We provided a detailed computational studies on a hydroxo-bridged diiron(III)-bisporphyrin
reported by Sankar and co-workers (Bhowmick. S, Ghosh, S.K, et al, 2012) who reported the
synthesis, structure and spectroscopic characterization of 1,2-bis[-hydroxo iron(III) 5-
(2,3,7,8,12,13,17,18-octaethylporphyrinyl)]ethane with I3–, BF4
– and ClO4
– counter anions.
Our computational modeling shows that as I3–, BF4
– and ClO4
– counter ion approach the
metal complex, the spin state ordering is affected dramatically. In order to understand how
the spin state ordering is affected by external perturbations, we have done a comprehensive
computational study. The calculations show that subtle environmental perturbations, such as
entropic corrections, perturb the spin state ordering and relative energies and are likely to be
the root cause of the variation in spin state ordering observed experimentally.
1
Mala A. Sainna, Debangsu Sil, Dipankar Sahoo, Bodo Martin, Sankar Prasad Rath, Peter Comba and Sam P. de Visser, “Spin state
ordering in μ-hydroxo bridged diiron(III)bisporphyrin complexes”. Inorganic Chemistry, 2015,54(4)1919-1930
81
3.1 Introduction
Spin state ordering is an important feature in transition metal catalyzed reactions in biology
as well as in synthetic models, where the spin state of the reactant often determines the
catalytic efficiency of the system. Nature often uses transition metals as catalytic reaction
centers in enzymes, and metal centers are also used as electron transfer complexes. In many
of these systems the active center has two (or more) interacting metal ions which often are
iron centers. Examples of diiron containing enzymes and proteins include hemerythrin
(Wilkins, P.; Wilkins, R. 1987, Stenkamp, R. 1994) , ribonucleotide reductase (RNR)
(Nordlund, P.; Reichard, P. 2006, Tomter, A. B.; Zoppellaro, G.; 2013, Krebs, C.; Dassama,
L. et al 2013), ferritin,( Moënne-Loccoz, P.; Krebs, C. et al 1999) methane monoxygenase
(MMO), (Murray, L., Lippard, S. 2007, Balasubramanian, R.; Rosenzweig, A. 2007) and 9-
stearoyl-acyl carrier protein desaturase (Shanklin, J.; Somerville, C. 1991,Fox, B., Lyle, K.
2004), to name a couple of well-studied systems. These diiron enzymes have diverse
functions in biology and all have the two iron ions separated by about 3 – 5 Å and linked to
the protein via nonheme interactions with histidine and carboxylate (Glu/Asp) residues.
Hemerythrin and ferritin are proteins involved in oxygen transport and iron storage, whereas
other diiron enzymes, such as RNR, catalyze the biosynthesis of deoxyribonucleotides from
ribonucleotides. MMO is one of the most efficient biochemical oxidants in nature and one of
the few enzymes that can hydroxylate methane.
By contrast to these nonheme diiron proteins, cytochrome c oxidase (CcO) has a central
dimetal complex, where one of the metals binds to a heme group and the other is in a
nonheme environment. CcO is a trans-membrane protein that reduces molecular oxygen to
water and thereby releases four electrons that are shuttled through the membrane (Ferguson-
82
Miller, S.; Babcock, G. 1996, Collman, J., Boulatov, R. 2004, Belevich, I.; Verkhovsky, M.
2006, Siegbahn, P.; Blomberg, M. 2008). To enable oxygen reduction it has a specific active
site structure with two metal centers in close proximity. The intricate details of dioxygen
binding and reduction as well as the electronic and spin state coupling between the two metal
centers in dimetal proteins, such as CcO, are still shrouded in mystery. In order to gain
insight into spin state interactions between metal centers, synthetic analogues have been
created that model the enzymatic systems (Collman, J, Boulatov, R. et al 2004, Chufán, E,
Puiu, S. et al 2007). These studies have given insight into the electronic and spectroscopic
properties of short-lived intermediates in the catalytic cycle of CcO and other dimetal enzyme
centers.
In several diiron enzymes, such as MMO and RNR, catalytic intermediates have been
proposed with either an oxo, peroxo or hydroxo group bridging the two iron centers,
frequently in conjunction with a bridging carboxylate group (Rosenzweig, A.; Lippard, S.
1994, Dassama, L.; Silakov, A. et al 2013, Korendovych, I.; Kryatov, S. et al 2007). The spin
multiplicity on each of the metal centers has an effect on the overall reaction mechanism and
may vary during the catalytic cycle. Thus, in many metal-catalyzed reactions a multistate
reactivity pattern has been established on low lying electronic and spin states (Shaik, S.;
Filatov, M. et al 1998, Shaik, S.; de Visser, S. et al 2002). For instance, computational
modeling established a degenerate pair of doublet and quartet spin states for the active
oxidant of cytochrome P450, i.e. compound I (Green, M. 1999, de Visser, S.; Kumar, D.
2011). It was found that in each of these spin states the oxidant can react with a substrate to
form products with different rate constants and sometimes even different mechanisms.
83
Understanding the factors that contribute to spin state ordering and relative energies in
transition metal complexes, therefore, is important and may affect reactivity patterns.
Oxo- and hydroxo-bridged diiron active centers are common structural motifs found among
proteins involved in O2 metabolism. The transformation of an oxo to a hydroxo bridge is a
proposed step in the reaction pathways of a great variety of iron and copper redox enzymes.
As enzymatic systems are difficult to study experimentally, synthetic model complexes have
been developed that contain the characteristic features of enzyme active sites. Recently,
Sankar and co-workers reported the spectroscopic characterization of a series of 1,2-bis[-
hydroxo iron(III) 5-(2,3,7,8,12,13,17,18-octaethylporphyrinyl)]ethane complexes, 1b•X
(Scheme 3.1; X = I3–, BF4
– and ClO4
–), in which the two iron(III) centers are either equivalent
or nonequivalent based on the counter anions present, although both cores have identical
molecular structures (Bhowmik, S.; Ghosh, S. et al 2012, Ghosh, S.; Rath, S. 2010). All
complexes were crystallographically characterized and showed considerable doming of both
porphyrin cores and out-of-plane displacement of the metal ions. Also, the spin states were
found to be dependent on the counter ions used. In particular, spectroscopic investigation of
1b•I3 identified core I (shown in red in Scheme 3.1) as having nearly high-spin configuration
(S = 5/2 with a minor contribution of S =
3/2), whereas core II (shown in blue in Scheme 3.1)
had an admixed intermediate-spin state (S = 3/2 with a minor contribution of S =
5/2) both in
the solid and solution phases (Bhowmik, S.; Ghosh, S. et al 2012, Ghosh, S.; Rath, S. 2010).
Complex 1b•BF4 was, however, found to have two iron(III) centers with admixed
intermediate states, whereas 1b•ClO4 was characterized to have high-spin (S = 5/2) iron(III)
centers in the solid but nearly intermediate (S = 3/2) spin in solution. It appears, therefore, that
changing the counter ion affects the spin state ordering and spin coupling between the two
84
iron centers. Dependence of the spin states of the metal complexes on counter anions has
been found rarely, particularly with anions which are far away and not apparently involved in
a direct interaction with the metal site (Bhowmik, S.; Ghosh, S. et al 2012, Ghosh, S.; Rath,
S. 2010, Ghosh, S.; Bhowmik, S. 2012, Kitchen, J.; White, N. et al 2011, Kitchen, J.;
Brooker, S. 2008, Nihei, M.; Shiga, T. et al 2007, Yamada, M.; Hagiwara, H. et al 2006,
Gütlich, P.; Hauser, A. et al 1994). How counter anions affect the spin state ordering and
coupling of metal centers is currently unknown and encouraged us to do a further
computational study.
In the present work, the synthesis, structure and properties of the corresponding μ-hydroxo
diiron(III)bisporphyrin complexes are reported here and compared with similar complexes
reported before. The ethyl linker, that bridges two porphyrin rings, introduces significant
horizontal and vertical flexibility to the bisporphyrin framework and also brings the two
porphyrin moieties close enough to distort the macrocycles significantly in the -hydroxo
complexes.
Scheme 3.1: Structures investigated in this work.
N N
N N
R R
R R
R
R
R
R Fe III
N N
N N
R R
R R
R R
R R
Fe III
O
N N
N N
R R
R R
R
R
R
R Fe III
N N
N N
R R
R R
R
R R
R Fe III
O H X OH
H X
1 X .
X : I 3 , BF 4 , ClO 4
1a : R = H
1b : R = C 2 H 5
85
3.2 Methods
The work described here is focused on the computational description of -hydroxo
diiron(III)bisporphyrin 1•X with particular focus on the spin state ordering and relative
energies. Due to the size of the system, some of the calculations were done on the full model
(with ethyl side chains on both porphyrin rings), 1b, whereas other calculations use an
abbreviated model with all side chains replaced by hydrogen atoms, 1a. The initial work was
done on the crystal structure coordinates of the -hydroxo-diiron(III)bisporphyrin 1b•I3 with
the counter anion removed, Scheme 3.1, with a total of 176 atoms and overall charge +1.
Data was benchmarked against the data from ref (Bhowmik, S.; Ghosh, S. et al 2012, Ghosh,
S.; K. Rath, S. 2010), using a range of density functional theory methods and basis sets and
initially focused on the spin state ordering and energies.
Technically, a bridged -hydroxo diiron(III)bisporphyrin can exist in three possible spin
configurations on each metal center that are either ferromagnetically or antiferromagnetically
coupled. As such, there are a large number of possible spin-spin interactions between the
metal centers of which we took a representative sample. Experimental studies on 1b•I3
indicate a spin multiplicity of S = 5/2 on the iron center of core I and a gross multiplicity of
3/2
on the iron center of core II (Bhowmik, S.; Ghosh, S. et al 2012, Ghosh, S.; Rath, S. 2010). In
a first set of calculations we took the crystal structure coordinates of 1b•I3. As we were
interested in the free energy differences of the various spin states we attempted a frequency
calculation on this structure. However, as the coordinates were taken from the crystal
structure, this had a large number of imaginary frequencies. We, therefore, ran a subsequent
constrained geometry optimization, whereby the coordinates of all ethyl side-chains were
86
minimized and the porphyrin cores kept fixed. This gave a frequency calculation with few
imaginary frequencies.
Structure 1b•I3 was calculated (without I3– counterion) with an overall charge of +1 and
total spin multiplicities of Stot (core I, core II) = 5 (5/2,
5/2), 4 (
5/2,
3/2), 3 (
3/2,
3/2) and 1 (½, ½).
All calculations were performed using the GAUSSIAN-09 suits of programs (Frisch, M.
2004) exploiting unrestricted density functional theory (DFT) methods with a variety of
hybrid and non-hybrid exchange-correlation functionals: B3LYP (Becke, A. 1993, Lee, C.;
Yang, W. et al 1988), B3LYP* (i.e. B3LYP with 15% HF exchange) (Reiher, M.; Salomon,
O. et al 2001), BP86 (Becke, A. 1988, Perdew, J. 1986), OLYP (Handy, N. C.; Cohen, A.
2001, Hoe, W.; Cohen, A. 2001), OPBE (Perdew, J.; Burke, K. 1996), and TPSSh (Tao, J.;
Perdew, J. et al 2003).
We tested three different basis set combinations labeled as BS1, BS2 and BS3. All initial
studies were done with a modest double- quality basis set on iron that contains an effective
core potential (LANL2DZ), in combination with a 6-31G basis set on the remaining atoms:
BS1 (Hay, P.; Wadt, W. 1985, Hehre, W.; Ditchfield, K. et al 1972). This basis set was
chosen to reduce computational cost as well as used as a comparison with a mixed valence
basis set of triple- quality consisting of DEF2-TZVP (Schaefer, A.; Huber, C. et al 1994) on
iron and the four nitrogen ligands in the first coordination sphere of each of the porphyrin
rings, whereas the remaining atoms were calculated using a 6-31G basis set: BS2. Finally, the
atoms of the complete complex 1b were also calculated with the triple- quality basis set
DEF2-TZVP throughout: BS3. The system was calculated with each of the basis sets BS1,
BS2 and BS3 coupled to all the density functional methods outlined above and the energetics
were tabulated for comparison.
87
In a final set of calculations, we ran several full geometry optimizations whereby the
counter anion was included. Calculations were performed on 1b•X and 1a•X, with X = I3–,
BF4– or ClO4
–, whereby all degrees of freedom were minimized. These calculations were run
in Jaguar (Schrödinger LLC. 2010) at the B3LYP/BS1 level of theory.
To understand the differences in spin state ordering between the individual porphyrin cores,
we also split structure 1b•I3 into individual cores, whereby each monomer was capped with a
hydrogen atom at the trimmed ethyl bridge and each of the separate cores were calculated
using the same BS1, BS2 and BS3 basis sets together with the chosen density functional
method. In these monomer units, we tested OH– and H2O as the distal ligand mimicking the
bridged -hydroxo group in structure 1b•I3. The two separated mono-porphyrins were also
capped with methoxy (CH3O–) or methanol (CH3OH) in place of OH
– and H2O, respectively,
because they may be geometrically closer to the actual conformation in 1b•I3.
3.3 Results and Discussion
To try to understand the origin of the spin state ordering experimentally characterized
hydro-bridged diiron porphyrin complexes, a series of computational studies on 1•X with
several counter anions were performed. Each core has quartet and sextet spin states that are
very close in energy, whereby the relative ordering is dependent on the environmental
perturbations as well as the nature of ligands and substituents. This has analogy to enzymatic
systems, and, for instance, in the resting state of cytochrome P450, i.e. the water bound
iron(III) heme cysteinate complex, the ground state is a doublet spin state (Thomann, H.;
Bernardo, M. et al 1995), but upon release of the water molecule the spin state ordering
changes and the sextet spin state becomes the ground state (Auclair, K.; Moënne-Loccoz, P.
88
et al 2001). In synthetic model complexes, such as cyanide ligated
iron(II)tetraphenylporphyrin, a rise of temperature leads to a spin crossover from low-spin to
high-spin (Li, J.; Lord, R. et al 2008). For [(py)2FeIII
(OEPO)] (OEPO = octaethyloxophlorin,
py = pyridine), the iron spin state is observed to be high-spin in solid which, however,
changes to a low-spin when dissolved in pyridine (Rath, S.; Olmstead, M. et al 2006).
Moreover, many cases have been reported on rate enhancements of oxygen atom transfer
reactions by axially ligated metal-porphyrin complexes (de Visser, S.; Ogliaro, F. et al 2002,
Nam, W.; Ryu, Y. et al 2004, Kumar, D.; de Visser, S. et al 2005, Song, W.; Ryu, Y. et al
2005, de Visser, S. 2006, de Visser, S.; Tahsini, L. et al 2009, Hessenauer-Ilicheva, N.;
Franke, A. et al 2009, Prokop, K.; de Visser, S. et al 2010, Takahashi, A.; Yamaki, D. et al
2012, Kumar, D.; Sastry, G. et al 2011).
In structure 1b•I3 there are two iron(III) centers, and both of these have three possible spin
states, which could give rise to 36 possibilities of coupling the spin states on core I with core
II. For instance, a spin of S = +5/2 on core I can couple with a spin of +
5/2, +
3/2, +
1/2, –
1/2, –
3/2
and –5/2 on core II. We did not test all these combinations, but decided to focus on the
ferromagnetic coupling between spin states with combinations (5/2,
5/2), (
5/2,
3/2), (
3/2,
3/2) and
(½, ½) only, which should give some idea on the relative stabilities of the resulting spin
states.
Figure 3.8 presents the spin state energies of the S = (5/2,
5/2), (
5/2,
3/2) and (
3/2,
3/2) states of
structure 1b as calculated with six different density functional methods. The S = (½, ½) states
were also calculated but found to be considerably higher in energy. We find dramatic
differences in spin state ordering and relative energies between the various DFT methods.
Thus, the experimental ground state of S = (5/2,
3/2) is only reproduced with the OPBE
89
method, although with both B3LYP and OLYP the S = (5/2,
3/2) and S = (
3/2,
3/2) states are
close in energy within 1.0 kcal mol–1
. Therefore, within the error of the calculations, the
B3LYP and OLYP results can be considered a good match with experiment. The BP86 and
B3LYP* methods, by contrast give large deviations for the experimentally predicted spin
state ordering and seem to be less suitable to predict the spin state ordering of -hydroxo
diiron(III)bisporphyrin complexes. To test whether the basis set has an influence on the spin
state ordering and relative energies, we ran single point calculations using a triple- quality
basis set on all atoms for structure 1b: OPBE/BS3. However, this calculation with a larger
basis set only gave minor changes in relative energies and did not change the ordering with
respect to OPBE/BS2.
Figure 3.1: Relative spin state energies of (5/2,
5/2), (
5/2,
3/2) and (
3/2,
3/2) states of structure 1b
as calculated with various DFT/BS2 methods in Gaussian. Calculations done with basis set
BS3 labeled with superscript a. (a) E values relative to the S = (5/2,
5/2) state. (b) E+ZPE
values relative to the S = (5/2,
5/2) state.
90
The results in Figure 3.8 show that changes in the density functional method affect the spin
state ordering and relative energies dramatically. In order to compare our work with
experimental data, zero point energies, thermal corrections, entropic and solvent corrections
were also included and each of these factors may have a different effect on the spin state
ordering and relative energies. Importantly, if a frequency calculation is run on crystal
structure coordinates of 1b, there are a large number of imaginary frequencies, mostly
connected to vibrations in the ethyl substituents. Therefore, a partial geometry optimization
with fixed cores and flexible ethyl side chains (distances, angles and dihedral angles
containing the C and H atoms of the ethyl groups) at the B3LYP/BS1 level of theory
preceded the frequency calculation. These structures with S = (5/2,
5/2), S = (
5/2,
3/2) and S =
(3/2,
3/2) spin states had no imaginary frequencies. ZPE corrections stabilize the S = (
5/2,
3/2)
state over the maximum spin state by 0.73 kcal mol–1
, whereas the S = (3/2,
3/2) state is
destabilized by 3.2 kcal mol–1
. Consequently, addition of ZPE corrections to the energies of
the data reported in Figure 3.8 improves the comparison between experiment and theory for
most methods, and now the correct spin state ordering is reproduced with B3LYP, OLYP,
TPSSh and OPBE. Despite the fact that the energy gap between the S = (5/2,
3/2) and the S =
(3/2,
3/2) states has narrowed considerably after addition of ZPE corrections, the BP86 and
B3LYP* methods still predict the wrong ground state.
We then investigated the effect of the zero-point energy, thermal and entropic corrections
on the spin state ordering and associated relative energies. It is found that free energy lowers
the S = (5/2,
3/2) state over the S = (
5/2,
5/2) state with respect to the E values by 1.4 kcal mol
–
1, while the S = (
3/2,
3/2) state is stabilized over the S = (
5/2,
5/2) state by 0.4 kcal mol
–1. This
91
implies that at a free energy level, the experimental spin state ordering is reproduced using
the B3LYP, OLYP and OPBE methods only.
To understand the origin of the large variation in spin state ordering and relative energies
between the various DFT methods we decided to split structure 1b into two individual cores
and calculate those separately. Thus, we took structure 1b and separated it into two cores,
whereby the bridging ethyl group was included in both cores and capped with a hydrogen
atom. We initially investigated both cores with a hydroxo ligand in the distal position and
Figure 3.9 displays the spin state ordering and relative energies of core I (panel a) and core II
(panel b), respectively. Thus, all DFT methods predict an S = 3/2 ground state for core II, but
the spin state ordering for core I varies with the DFT methods. Moreover, the quartet-sextet
energy splitting varies by more than 13 kcal mol–1
for each individual core, whereby the
BP86 method gives a large stabilization of the quartet spin state over the sextet spin state.
Hence, the over-stabilization of lower spin states by pure density functionals such as BP86,
therefore, makes these methods unsuitable for the correct description of -hydroxo
diiron(III)bisporphyrin complexes. Also the reduced amount of HF exchange in the B3LYP*
method over-stabilizes the quartet spin states by a too large amount and therefore gives the
wrong spin state ordering of the complex. The B3LYP, OLYP, TPSSh and OPBE methods all
predict the correct spin state ordering for each of the individual cores and give a small
quartet/sextet spin state energy gap.
92
Figure 3.2: Quartet/sextet spin state energies for (A) individual core I and (B) core II. All
energies are relative to the quartet spin state in kcal mol–1
. A negative value implies an S = 5/2
ground state.
A comparison of the quartet-sextet energy splitting in core I as compared to the energy
difference in 1b between the S = (5/2,
3/2) and S = (
3/2,
3/2) states is a measure for the
intramolecular interactions of core II on core I. In all cases, the dinuclear complex shows
extra stabilization of the S = 3/2 state on both core I and core II. On core I the quartet spin
state of the isolated core with respect to the dinuclear complex is stabilized over the sextet
spin state by 1.8 – 3.7 kcal mol–1
for the seven methods described in Figures 3.8 and 3.9,
whereas a stabilization of 2.1 – 6.2 kcal mol–1
is found for core II. This means that the
antiferromagnetic interaction between the two cores, the porphyrin ring deformation and -
stacking interactions lead to a stabilization of the quartet spin state on each core with respect
to the sextet spin state, which is independent of the density functional method and basis set.
Ring-deformation is known to stabilize intermediate spin states over high-spin states in
iron(III)-porphyrin complexes (Bhowmik, S.; Ghosh, S. et al 2012, Ghosh, S.; Rath, S. et al
2010, Ghosh, S.; Bhowmik, S. et al 2013). Indeed, in our particular system both porphyrin
93
rings show considerable ruffling, but not by the same degree, hence leading to different spin
states on each iron center.
To test the effect of ZPE, thermal and entropic corrections on the spin state ordering of the
two individual cores, we ran a partial geometry optimization at B3LYP/BS1, where we fixed
the first and second coordination sphere of atoms around the iron center and reoptimized the
rest. This led to a structure without imaginary frequencies and almost negligible change in
ZPE for both spin states: The quartet spin state is stabilized over the sextet spin state by 0.2
and 0.1 kcal mol–1
for core I and II, respectively. Subsequently, we investigated the free
energy differences of these structures in the relevant spin states, and when the full
combination of ZPE, thermal and entropic corrections are included the sextet spin state is
favored by 0.9 and 0.6 kcal mol–1
for core I and core II, respectively. This is not surprising as
the entropy component contains a factor –RT ln (2S +1), which has a value of 4.4 kcal mol–1
for the sextet spin state and 3.4 kcal mol–1
for the quartet spin state at a temperature of 298 K.
As our spin state splittings are small, this extra stabilization of the free energy due to electron
spin may tip the balance in favor of the high-spin state.
We did a further set of test calculations on the individual cores I and II, where we replaced
the OH ligand by a water molecule. However, with all methods (except B3LYP*) a quartet
spin ground state was found for both core I and core II. This again shows that subtle changes
to the coordination system of iron(III)porphyrins can lead to significant changes in spin state
ordering and relative energies. To find out whether the difference between an OH ligand
versus a water ligand is an effect of the charge on the metal center, we also calculated the
same complexes with CH3O– and CH3OH bound in the distal ligand position. At OPBE/BS2
level of theory with a CH3O– ligand, we find core I to have a sextet spin ground state by 4.6
94
kcal mol–1
, whereas core II has a quartet spin ground state by 2.0 kcal mol–1
. This spin state
ordering match those found for OH– as well as those observed experimentally. By contrast,
when we replace methoxide by methanol and redo these calculations at the same level of
theory, we find core I to have a quartet spin ground state by 2.6 kcal mol–1
, and core II has a
sextet spin ground state by 16.9 kcal mol–1
. In summary, binding of an anionic ligand to an
iron(III) center affects the spin state ordering with respect to binding a neutral (solvent)
molecule to an iron(III) center.
To further establish the environmental effects on the spin state ordering and relative
energies, we decided to include counter anions, i.e. I3–, BF4
– and ClO4
–, into the model. We
took the crystal structure coordinates of 1b•X with X = I3–, BF4
– and ClO4
– and did a full
geometry optimization without constraints for the three relevant spin states, i.e. S = (5/2,
5/2), S
= (5/2,
3/2) and S = (
3/2,
3/2), for 1a•X and 1b•X. The obtained structures of the complexes are
shown in Figure 3.10. Counter ions have small and virtually insignificant effects on the
optimized geometries and, as follows from Figure 3.10, we find very similar structures for
1a•X (X = I3–, BF4
– or ClO4
–). The same trends are found for the large system with ethyl
substituents attached to the porphyrin scaffold (1b•X). Of course, there are geometric
differences between the spin states as discussed above. A geometry optimization elongates
both Fe–O distances from 1.897/1.934 Å in the crystal structure to slightly over 2.0 Å. On the
other hand, very little changes are found for the porphyrin groups: the average Fe–N
distances in core I/core II are 2.051/2.006 Å in the crystal structure, which are values close to
those seen in Figure 3.10. Also the displacement of the metal from the plane through the four
nitrogen atoms (defined as ) is virtually the same in the optimized geometries as compared
to the crystal structures.
95
Figure 3.3: Optimized geometries of the S = (5/2,
5/2), [S = (
5/2,
3/2)] and {S = (
3/2,
3/2)} states
of 1a•X and 1b•X with X = I3–, BF4
– or ClO4
– at the B3LYP/BS1 level of theory in Jaguar.
Bond lengths are given in angstroms and is the average deviation of the iron from the plane
through the four nitrogen atoms.
Table 3.1 provides relative energies of the S = (5/2,
5/2), S = (
5/2,
3/2) and S = (
3/2,
3/2)
optimized geometries of 1a•X and 1b•X (X = I3–, BF4
– or ClO4
–) as calculated with B3LYP
and OPBE. Energetically, in all cases the S = (3/2,
3/2) state is the ground state followed by the
S = (5/2,
3/2) state and the S = (
5/2,
5/2) state. Only very small changes in energies are obtained
when counter ions are added and no clear changes are observed, based on the nature of the
counter ions. The same trends are found for B3LYP and OPBE as density functional method.
Clearly, in the gas phase the effect of counter ions is small and does not induce changes in the
spin state ordering and relative energies. As discussed above, environmental effects can
rFeO = 2.013 [2.080] {2.050}
rFeO = 2.004 [1.978] {2.041}
rFeN,average = 2.071 [1.992] {1.998} = 0.456 [0.274] {0.300}
rFeN,average = 2.073 [2.080] {2.004} = 0.433 [0.458] {0.287}
rFeO = 2.020 [1.992] {2.006}
rFeO = 2.018 [2.084] {1.965}
rFeN,average = 2.074 [2.083] {2.073} = 0.453 [0.482] {0.447}
rFeN,average = 2.075 [2.000] {2.001} = 0.446 [0.267] {0.256}
rFeO = 2.006 [2.072] {2.042}
rFeO = 2.014 [1.988] {2.052}
rFeN,average = 2.073 [1.990] {1.999} = 0.441 [0.264] {0.291}
rFeN,average = 2.073 [2.076] {1.996} = 0.455 [0.470] {0.296}
rFeO = 2.018 [1.988] {2.056}rFeO = 2.014 [2.086] {2.052}
rFeN,average = 2.086 [2.088] {2.011} = 0.450 [0.490] {0.323}
rFeN,average = 2.083 [2.003] {2.013} = 0.458 [0.284] {0.309}
rFeO = 2.001 [2.079] {1.941}rFeO = 2.001 [1.975] {1.996}
rFeN,average = 2.068 [1.975] {1.976} = 0.479 [0.285] {0.275}
rFeN,average = 2.069 [2.078] {2.073} = 0.425 [0.469] {0.444}
rFeO = 2.005 [1.981] {2.055}
rFeO = 1.996 [2.065] {2.041}
rFeN,average = 2.065 [2.072] {1.984} = 0.470 [0.492] {0.314}
rFeN,average = 2.065 [1.982] {1.990} = 0.424 [0.270] {0.299}
96
change the relative spin state ordering and in some cases a different electronic ground state is
observed experimentally in solution than in a crystal structure.
Table 3.1. Relative energies of optimized geometries in different spin states of 1a•X and
1b•X complexes.a
a Energies in kcal mol
–1 with ZPE corrected values in parenthesis.
E (E+ZPE) E (E+ZPE) E (E+ZPE) E (E+ZPE) E (E+ZPE)
1aX X = I3– X = BF4
– X = ClO4
–
B3LYP/BS1 B3LYP/BS1 OPBE/BS1 B3LYP/BS1 OPBE/BS1
S = (5/2,
5/2) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00)
S = (5/2,
3/2) –6.07 (–5.23) –4.99 (–3.88) –5.83 (–4.70)
S = (3/2,
3/2) –9.87 (–7.79) –9.83 (–7.78)
1bX
S = (5/2,
5/2) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00)
S = (5/2,
3/2) –5.90 (–4.84) –6.11 (–5.16) –2.83 (–1.66) –5.69 (–4.75) –3.00 (–1.61)
S = (3/2,
3/2) –11.08 (–8.74) –4.62 (–2.29) –9.12 (–7.09) –4.29 (–1.53)
97
Interestingly, the relative energies in Table 3.3 are similar to those reported above in Figure
3.10a, where we reported single point DFT calculations on the crystal structure coordinates.
However, correction for ZPE by a partial geometry optimization lowered the S = (5/2,
3/2)
below the S = (3/2,
3/2) state, which is not seen when a full geometry optimization and
frequency calculation is performed. As such, minor geometric distortions to the structure
affect the spin state ordering and electronic ground state of -hydroxo-bridged diiron
complexes. We should note here that the relative energies can be further affected by spin-
orbital coupling also (Gupta, A.; Gupta, G. et al 1994, Hagen, W.; van den Berg, W. et al
1998). It may very well be that the degree of spin-orbit coupling is affected by nearby anions,
such as counter anions, and, thereby changes the spin state ordering as observed
experimentally.
The data shown in Table 3.3 indicate that the energetic differences between the three main
spin states are very small and that in the gas phase, there is a preference for an S = (3/2,
3/2)
spin state. As reasoned above, entropic corrections will favor higher spin states and
consequently, an increase of the temperature will strongly influence the spin state ordering.
We note, however, that in the optimized geometries, the counter anion is aligned with the
dipole moment vector of the diiron porphyrin complex as expected for an ion-dipole
interaction. The change in spin state upon addition of counter ions observed experimentally,
may, therefore, result from a combination of factors, of which some are not included in the
computational models. The hydrogen bonding interactions of solvent molecules to the
bridging -hydroxo group may play a key role in establishing the electronic ground state, and
this was not considered in the models used here. Furthermore, an induced electric field was
shown to strongly affect the spin state ordering, which may have an effect in the systems
98
studied here (Shaik, S.; de Visser, S. et al 2004, de Visser, S. 2005). Finally, spin-orbit
coupling can influence the small energy differences and change the relative ordering between
the various spin states.
It would be interesting to compare the spin states of µ-hydroxo diiron(III)bisporphyrins
obtained out of experiment and theory. Experimentally, 1b•I3 is assigned to have nearly high-
spin and admixed-intermediate spins of iron in cores I and II, respectively, which is also
reproduced in theory using the crystal structure coordinates of the molecule utilizing a
variety of hybrid and non-hybrid exchange correlation functions. Energetically, the S = (3/2,
3/2) state is the ground state followed by the S = (
5/2,
3/2) state and then the S = (
5/2,
5/2) state
which happens to be the correct spin state ordering obtained experimentally for most of the
complexes described here. It is interesting to note here that two Fe centers in 1b•ClO4 which
are found to be high-spin (S = 5/2) in the solid state converts to the intermediate-spin (S =
3/2)
in solution. One probable reason could be the breaking of H-bonding interactions observed in
the solid once it dissolved in solution. Thus, subtle change in the environment can also affect
the relative spin state ordering and in some cases a different electronic ground state. The
relative energies of the spin states are also known to be affected by the extent of spin-orbital
coupling which can be influenced by the nearby counter-anions. Moreover, an induced
electric field can also affect spin state ordering and relative energies therein. Thus, the
different spin-state behavior of 1b•X, as observed in the experiment, can be attributed to the
properties of the counter anion X which are also known to operate under the influence of
variety of phenomena related to steric effects, charge polarization, stability of the ion-pair
formations etc. and were not considered in the computational models used here.
99
3.4 Conclusion
I have provided a detailed computational studies on a diiron(III)bisporphyrin complex with
particularly focus on the spin state ordering. An earlier experimental work by Sankar and co-
workers shows a strong variation in electronic ground state for 1b•X (X = I3–, BF4
–, ClO4
–)
(Bhowmick. S.; Ghosh. S. et al, 2012). Crystallographic studies show considerable ruffling of
both porphyrin rings by unequal amounts, and the doming and structural features are
characteristic of a spin state. 1H NMR and Mössbauer studies confirmed this assignment.
Finally, an extensive computational study was performed on various diiron(III)bisporphyrin
complexes with a range of DFT methods, which established a variety of factors that affect
spin state ordering and relative energies in -hydroxo bridged diiron(III)bisporphyrin
complexes. The latter highlights the small energy gap between the S = (3/2,
3/2) and S = (
5/2,
3/2) spin states and how their ordering can interchange through external perturbations.
100
CHAPTER 4
PROJECT TWO
101
A comprehensive test set of epoxidation rate
constants by iron(IV)-oxo porphyrin complexes2
ABSTRACT
Cytochrome P450 enzymes are heme based monoxygenases that catalyse a range of oxygen
atom transfer reactions with various substrates, including aliphatic and aromatic
hydroxylation as well as epoxidation reactions. The active species is short-lived and difficult
to trap and characterize experimentally, moreover, it reacts in a regioselective manner with
substrates leading to aliphatic hydroxylation and epoxidation products, but the origin of this
regioselectivity is poorly understood. Crestoni and co-worker’s synthesized a model complex
and studied it with low-pressure Fourier transform-ion cyclotron resonance (FT-ICR) mass
spectrometry (MS). A novel approach was devised using the reaction of [FeIII
(TPFPP)]+
(TPFPP = meso-tetrakis(pentafluorophenyl)porphinato dianion) with iodosylbenzene as a
terminal oxidant which leads to the production of ions corresponding to [FeIV
(O)(TPFPP+•
)]+.
This species was isolated in the gas-phase and studied in its reactivity with a variety of
olefins. Product patterns and rate constants under Ideal Gas conditions were determined by
FT-ICR MS. All substrates react with [FeIV
(O)(TPFPP+•
)]+ by a more or less efficient oxygen
atom transfer process. In addition, substrates with low ionization energies react by a charge-
transfer channel, which enabled us to determine the electron affinity of [FeIV
(O)(TPFPP+•
)]+
2 Mala A. Sainna, Suresh Kumar, Devesh Kumar, Simonetta Fornarini, Maria Elisa Crestoni and Sam P. de Visser, “A comprehensive test
set of epoxidation rate constants by iron(IV)-oxo porphyrin complexes”. Chemical Science, 2015, 6, 1516-1529.
102
for the first time. Interestingly, no hydrogen atom abstraction pathways are observed for the
reaction of [FeIV
(O)(TPFPP+•
)]+ with prototypical olefins such as propene, cyclohexene and
cyclohexadiene, which suggests that the competition between epoxidation and hydroxylation
– in the gas-phase – is in favour of substrate epoxidation. This notion further implies that
P450 enzymes will need to adapt their substrate binding pocket, in order to enable favourable
aliphatic hydroxylation over double bond epoxidation pathways. The MS studies yield a large
test-set of experimental reaction rates of iron(IV)-oxo porphyrin complexes, so far
unprecedented in the gas-phase, providing a benchmark for calibration studies using
computational techniques. My computational results presented here confirm the observed
trends excellently and rationalize the reactivities within the framework of thermochemical
considerations and valence bond schemes.
4.1 Introduction
The cytochromes P450 are part of the body’s natural defence mechanism in the liver and
perform vital functions for human health that include the biodegradation of xenobiotic and
drug molecules (Sono, M.; Roach, P. et al 1996, Groves, T. 2003, Ortiz de Montellano, R.
2004). Due to this broad chemical function the P450s can bind and activate a large range of
substrates with varying shapes and sizes. Generally, the P450s act as monoxygenases,
whereby they bind and utilize molecular oxygen via a heme centre and transfer one of the
oxygen atoms of O2 to a substrate, while the second oxygen atom leaves the process as a
water molecule. The P450s react with substrates activating aliphatic and aromatic
hydroxylation, epoxidation and sulfoxidation reactions, but have also been reported to
catalyse desaturation and N-dealkylation reactions (Groves, T. 2005, Watanabe. Y.;
103
Nakajima, H.; et al 2007). There are many different P450 isozymes and until early 2014
thousands of different structures had been characterized (Nelson, R. 2006). All P450s share
common features which include a catalytically active heme group with a central iron atom
that is linked to the protein by the thiolate sulphur atom of a cysteinate side chain (Poulos, L,
Finzel, C. et al 1985, Poulos, L.; Finzel, C. et al 1986). Fig 4.1 displays the structure of a
typical P450 active site, namely the one belonging to the CYP124 isozyme as taken from the
2WM4 protein databank (pdb) file (Johnston, B.; Kells, M. et al 2009). As shown in Fig 4.1
the substrate (tyramine) is located in a cleft nearby the heme, the substrate binding pocket,
which is in a tight orientation with stabilizing hydrogen bonding interactions by several
residues. The vacant sixth coordination site of iron is the position where molecular oxygen
will bind during the catalytic cycle. The process includes two reduction and two protonation
steps to synthesize the active species of P450 called Compound I (Cpd I) (Meunier, B.; de
Visser, P. et al 2004, de Visser, P.; Kumar, D. 2010). Cpd I is highly reactive and therefore
difficult to study experimentally, however, a few reports on its spectroscopic properties have
appeared in the literature (J. Rittle, J.; Green, T. 2010).
104
Figure 4.1: Active site of P450 as taken from the 2WM4 pdb file.
Due to the short lifetime of Cpd I, studying catalytic mechanisms and reaction rates of P450
catalysed reactions is challenging (Auclair, K.; Hu, Z. et al 2002, Lafite, P.; André, F. et al
2007, Dowers, S.; Rock, A. et al 2004, Cooper, R.; Groves, T. 2011, Roberts, M.; Jones, P.
2010); therefore, many studies have focused on biomimetic iron-porphyrin complexes instead
(Groves, T.; Myers, S. 1983, Groves, T.; Watanabe, Y. 1986, Ostović, D.; Bruice, C. 1989,
Groves, T.; Gross, Z. 1994, Stephenson, N.; Bell, T. 2006, Hessenauer-Ilicheva, N.; Franke,
A. et al 2007, Mas-Balleste, R.; Que Jr, L. 2007, Comba, P.; Rajaraman, G. 2008, Bruijnincx,
A.; Buurmans, C. et al 2008, Hull, F.; Sauer, O. et al 2009, McGown, A.; Kerber, D. et al
2009, Company, A.; Feng, Y. et al 2009, Franke, A.; Wolak, M. et al 2009, Hessenauer-
Ilicheva, N.; Franke, A. et al 2009). These studies gave detailed insight into the effect of axial
and equatorial ligands (Gross, Z.; Nimri, S. 1994, Gross, Z. 1996, Czarnecki, K.; Nimri, S. et
al 1996), but also on the local environment such as the substrate binding pocket. A
105
particularly useful method to establish the properties and reactivity patterns of short-lived
complexes, such as catalytic intermediates, is Fourier transform – ion cyclotron resonance
(FT-ICR) mass spectrometry (MS) (Marshall, G.; Hendrickson, L. et al 1996, Roithová, J.;
Schröder, D. 2010, Nibbering, M. 2006). In FT-ICR MS, the charged species of interest
(either positive or negative ion) is trapped in a collision cell for a specific time during which
reactions with neutral gases can occur and be studied at the prevailing low pressure of the
instrument. FT-ICR MS allows one to measure the ion distributions and fragmentation
patterns at varying trapping time, thereby yielding insight into reactivities, and enabling one
to calculate rate constants and thermochemical properties. In recent work, Crestoni, Fornarini
and co-workers have succeeded in trapping and characterizing the Cpd I analogues of iron
and manganese porphyrin complexes and studied their reactivity with a selection of substrates
(Crestoni, E.; Fornarini, S. 2005, Crestoni, E.; Fornarini, S. 2007, Crestoni, E.; Fornarini, S.
et al 2010, Lanucara, F.; Crestoni, E. 2011). Thus, the [MnV(O)(TPFPP)]+ complex (TPFPP
= meso-tetrakis (pentafluorophenyl)porphinato dianion) was found to react with model
substrates through oxygen atom transfer (OAT), electron transfer (ET), hydride transfer and
ligand addition. However, no direct hydrogen atom transfer (HAT) with any tested substrate
took place.
In order to find out what drives the OAT reaction of Cpd I with olefins, Crestoni and co-
workers decided to investigate the properties and reactivities of [FeIV(O)(Por+•
)]+ (Por =
porphine dianion) and [FeIV(O)(TPFPP+•
)]+ with FT-ICR MS and relate the findings with
density functional theory (DFT) methods I did in order to rationalize the trends. These studies
represent the first comprehensive computational – study benchmarked against experiment on
olefin epoxidation by iron(IV)-oxo porphyrin cation radical models and allow correlations to
106
be established between the OAT rate constant and the ionization energy (IE) of the olefin.
These correlations are further supported and rationalized by computational modelling.
Scheme 4.1: Models investigated in this work.
4.2 Methods
4.2.4 COMPUTATION
All calculations discussed here utilize density functional theory (DFT) methods as
implemented in the Jaguar and Gaussian-09 program packages (Schrodinger, 2011, Frisch, J.
2004). Two different models were investigated: (i) [FeIV
(O)(Por+•
)]+ (A) that includes a
porphyrin (Por) ring with all side-chains abbreviated to hydrogen atoms, and (ii)
[FeIV
(O)(TPFPP+•
)]+ (B), Scheme 4.1. Similar to previous work of ours in the field (Latifi, R.;
Sainna, A. et al 2013, Quesne, G.; Latifi, R. et al 2014), we use the unrestricted hybrid
density functional method UB3LYP19 as it was shown to reproduce the kinetics of
metal(IV)-oxo oxidants well (Vardhaman, K.; Sastri, V. et al 2011, Vardhaman, K.; Barman,
P. et al 2013, Kumar, S.; Faponle, S. et al 2014). Initial exploratory calculations employed a
107
modest LANL2DZ basis set on iron and 6-31G on the rest of the atoms (basis set BS1) (Hay,
J.; Wadt, R. 1985) for geometry optimizations, analytical frequencies and geometry scans.
These studies explored the potential energy surfaces involving reactants, intermediates and
products on different spin states in detail and generated starting structures for the transition
state optimizations. All local minima reported here had real frequencies only and the
transition states were characterized by a single imaginary frequency for the correct mode. To
improve the energetics of these structures we did single point calculations in the gas-phase
with a triple- quality basis set on iron (LACV3P+) and 6-311+G* on the rest of the atoms,
basis set BS2. Subsequently, all geometries (local minima and transition states) were
reoptimized at the UB3LYP/BS2 and UB3LYP-D3/BS2 levels of theory (Becke, D. 1993,
Lee, C.; Yang, W. 1988) (Grimme, S.; Antony, J. et al 2010) and characterized by an
analytical frequency analysis. Barrier heights reported in this work were calculated relative to
isolated reactants, although using reactant complexes instead only minor changes are
observed. The effect of solvent on the rate constants was tested through single point
calculations using the self-consistent reactant field model as implemented in Gaussian with a
dielectric constant representing chloroform ( = 4.7113).
Ionization energies and bond dissociation energy (BDEOH) values were calculated as
before (de Visser, P. 2010) and represent adiabatic values for reaction 1 and 2, respectively
and report UB3LYP/BS2//UB3LYP/BS1 energies including ZPE and dispersion corrections.
A A+•
+ e– + IEA (1)
[FeIV
(OH)(Por)]+ [Fe
IV(O)(Por
+•)]
+ + H
• + BDEOH (2)
108
4.3 Results
Scheme 4.2: Substrates investigated in this work.
109
a Ionization energies (IE, eV) are from ref (Lias, G.; Bartmess, E. et al) N/A stands for not
available. b Second-order rate constants (kexp) in units of 10
–10 cm
3 molecule
–1 s
–1 are
measured at a temperature of 300 K in the FT-ICR cell. The estimated error in kexp is 30%,
although the internal consistency of the data is within 10%. c Collision rate constants (kADO)
evaluated with the parameterized trajectory theory. d Reaction efficiency (%), = kexp/kADO ×
___________________________________________________________________________
Table 4.1. Kinetic data and product distributions obtained for the gas phase reaction of
[FeIV
(O)(TPFPP+•
)]+ with selected olefins as determined by FT -ICR MS.
___________________________________________________________________________
Substrate IE a kexp
b kADO
c
d HT CT OAT Add
ethene 10.51 8.5 10–5
8.5 1 10–3
– – 100 –
propene 9.73 7.6 10–3
9.45 0.080 – – 100 –
1-butene 9.55 0.029 9.6 0.30 – – 100 –
E-2-butene 9.10 0.080 10.8 0.74 – – 100 –
2,3-dimethyl-1-butene9.07 0.145 9.5 1.5 – – 100 –
cyclohexene e 8.95 0.194–0.291 9.7 2–3 – – 75 25
1,4-cyclohexadiene 8.82 0.511 9.29 5.5 – – 90 10
2-methoxy-1-propene8.64 0.819 10.5 7.8 – – 100 –
1,3-pentadiene 8.60 0.826 9.6 8.6 – – 100 –
styrene 8.46 1.40 9.26 15 – – 100 –
1,3-cyclohexadiene 8.25 1.58 9.29 17 – – 100 –
trans--methylstyrene 8.1–8.2 2.97 11.9 25 4 – 96 –
indene 8.14 3.18 8.6 37 2 12 86 –
-pinene f N/A 4.32–4.7 9.4 46–50 – – 100 –
110
100. e The reaction with cyclohexene-d10 gave a rate constant within experimental error of
that for cyclohexene-h10. f The IE for -pinene is 8.07 eV.
Scheme 4.3: Pathways observed for the reaction of [FeIV
(O)(TPFPP+•
)]+ ions (R = C6F5) with
selected substrates (Sub) as studied with FT-ICR MS.
Figure 4.2: Time dependence of relative ion abundancies for the reaction of
[FeIV
(O)(TPFPP+•
)]+ (m/z 1044) with indene. Product ions are [Fe
III(TPFPP)]
+ (m/z 1028),
0
20
40
60
80
100
0 5 10 15 20 25 30 35
m/z 116
m/z 1028
m/z 1044
m/z 1160
Abundance
[%]
time [seconds]
111
[Fe(TPFPP)(C9H8)O]+ (m/z 1160) and C9H8
+• (m/z 116). Experiments were performed in the
presence of indene at 5.2 10–8
mbar in the FT-ICR cell.
4.4 Theoretically derived reaction paths, energetics and structures.
The experimental studies reported above present a comprehensive test set of model reactions
of iron(IV)-oxo porphyrins with olefins for the first time and enable extensive benchmarking
and calibration of computational methods and procedures against gas-phase (Ideal Gas
conditions) rate constants. We decided to take the opportunity and calibrate previously used
methods and procedures for DFT studies on these chemical systems and compare to the
results of the FT-ICR rates from Table 4.1. In addition, the computational studies were
performed to further understand the substrate activation patterns by [FeIV
(O)(TPFPP+•
)]+ with
olefins, and rationalize the obtained trends. Before we will discuss details of the reaction
mechanism and possible reactivity trends, let us start with a detailed analysis of the reactant
species, namely [FeIV
(O)(Por+•
)]+,
4,2A, and [Fe
IV(O)(TPFPP
+•)]
+,
4,2B.
Fig 4.3 displays the high-lying occupied and low-lying virtual orbitals of 4,2
A; the orbitals
for 4,2
B look very similar. These orbitals are dominated by the interactions of the metal 3d
orbitals with its ligands and several -type porphyrin orbitals. Lowest in energy are a pair of
-type orbitals (z2 and xy): z
2 represents the -interactions of the 3dz
2 orbital on iron
with the 2pz orbital on oxygen, whereas the xy gives the interactions of the 3dxy orbital on
iron with 2px,y orbitals on the four nitrogen atoms of the porphyrin ligand. The antibonding
combinations of these two orbitals (*z2 and *xy) are high in energy and virtual. Also
doubly occupied is the x2–y
2 orbital, which is a lone-pair orbital located in the plane of the
porphyrin ring. Finally, the interaction of the metal 3dxz/3dyz with the 2px/2py on the
112
oxygen atom leads to a pair of xz/yz and a pair of *xz/*yz set of orbitals. The z2, xy,
xz and yz bonding orbitals are doubly occupied and low-lying in all calculations reported
here. In addition to the metal-type orbitals there are also two porphyrin-type -orbitals that in
D4h symmetry have the labels a1u and a2u. With a thiolate as axial ligand the a2u orbital
strongly mixes with a 3pz orbital on sulphur and hence is destabilized in energy (Green, T.
1999, Ogliaro, F.; de Visser, P. et al 2001), which strongly affects the electron affinity of the
oxidant and consequently is responsible for its push-effect (Dawson, H.; Holm, H. et al
1976).
Figure 4.3: Molecular valence orbitals of 4A.
2*z
xy*
ua2ua1
xz* yz*
yzxz
2z xy 22 yx
113
a Relative energies in kcal mol
–1 with respect to the
2A1u state, ND stands for not determined.
b Energies obtained at UB3LYP/BS2//UB3LYP/BS1 level of theory.
c Energies and
geometries calculated at UB3LYP/BS2 level of theory. d Energies and geometries calculated
at UB3LYP-D3/BS2 level of theory.
The set of orbitals displayed in Fig 4.3 is occupied with 15 electrons and as several of these
orbitals are close in energy there a number of possibilities to distribute the electrons over the
orbitals. In addition, states can also be found in various spin states ranging from doublet to
quartet and sextet, where we identify the spin state with a superscript in front of the electronic
state label. Thus, the electronic state labelled as 4A2u has an overall quartet spin state and
___________________________________________________________________________
Table 4. 2 . Relative energies of several low - lying electronic states of [Fe(O) (Por +•
)] + (A). a
___________________________________________________________________________
State Configuration A b A
c A d
E+ZPE E+ZPE E+ZPE
2 A 1u 2 * xz * yz
a 1u 0.00 0.00 0.00
4 A 1u 2 * xz * yz
a 1u 0.71 0.19 0.21
2 A 2u 2 * xz * yz
a 2u 1.65 3 .75 3.80
4 A 2u 2 * xz * yz
a 2u 1.25 3.42 3.47
6 A 2u * xz
* yz * xy
a 2u 9.25 18.68 19.38
4 xy * xz * yz
* xy a 1u
9.69 ND ND
4 zz * xz
* yz * zz
a 2u 19.53 ND ND
114
singly occupied a2u molecular orbital with overall electronic configuration: z22 xy
2 xz
2 yz
2
x2–y22 *xz
1 *yz
1 a1u
2 a2u
1 or in short [core] x2–y2
2 *xz
1 *yz
1 a1u
2 a2u
1. Similarly, we
calculated the doublet spin state (2A2u state), where the unpaired electron in the a2u orbital is
antiferromagnetically coupled to the unpaired electrons in the two * orbitals: 2A2u = [core]
x2–y22 *xz *yz a1u
2 a2u.
Previous studies with either imidazole, acetonitrile or thiolate as axial ligand (Green, T.
1999, Ogliaro, F.; de Visser, P. et al 2001, de Visser, P.; Shaik, S. et al 2003, Kamachi, T.;
Yoshizawa, K. 2003, Bathelt, M.; Zurek, J. et al 2005, de Visser, P. 2005, de Visser, P.;
Tahsini, L. et al 2009, Lonsdale, R.; Oláh, J. et al 2011, Isobe, H.; Yamanaka, S. et al 2011)
showed the 4,2
A2u states to be close in energy and well below alternative states. However, this
was due to considerable mixing of the a2u orbital with the axial ligand, which obviously is not
possible in our chemical system that lacks an axial ligand. However, in an isolated porphyrin
macrocycle, the a1u and a2u orbitals are degenerate (Ghosh, A. 1998); therefore, we decided to
investigate a range of possible electronic states for the pentacoordinated iron(IV)-oxo
porphyrin cation radical system, [FeIV
(O)(Por+•
)]+. Firstly, we tested the stability of the
4,2A2u
states and the alternative 4,2
A1u states with [core] x2–y22 *xz
1 *yz
a1u
1 a2u
2 orbital
occupation. In addition, we attempted to generate models with the iron in oxidation state
iron(V), i.e. 2xz state with occupation [core] x2–y2
2 *xz
1 a1u
2 a2u
2, or the iron in oxidation
state iron(III), i.e. the 4A state with orbital occupation [core] x2–y2
2 *xz
2 *yz
1 a1u
1 a2u
1 and
the 6xy,III state with [core] x2–y22 *xz
1 *yz
1 *xy
1 a1u
1 a2u
1 occupation. However, all our
attempts to calculate iron(III) or iron(V) states failed and converged back to lower lying
solutions with four un-paired electrons on the metal in a formal iron(IV) oxidation state,
115
hence the 2xz,
4A and 6xy,III states are high in energy and inaccessible to our chemical
system.
Table 4.2 summarizes relative energies of optimized geometries of the various electronic
spin states as calculated with different DFT methods for 2,4,6
A. As follows from Table 4.2 all
calculations of A give a 2A1u ground state that is nearly degenerate with the corresponding
quartet spin state. In general, calculations done at UB3LYP/BS2 and UB3LYP-D3/BS2 give
almost identical spin state orderings and relative energies, which shows that dispersion is not
a critical component for these chemical structures. Nevertheless, the 4,2
A2u and 4,2
A1u states
are close in energy and all four states could have a finite lifetime.
Optimized geometries of the 4,2
A2u and 4,2
A1u states are given in Fig 4.4. Geometrically, no
dramatic changes in bond lengths are obtained between the three optimization techniques. A
small basis set gives slightly longer Fe–O distances than those found with a triple- basis set.
The effect of dispersion is negligible on the optimized geometries: UB3LYP/BS2 and
UB3LYP-D3/BS2 give virtually the same chemical structures. Addition of meso-substituents
to the porphyrin ring such as pentafluorophenyl groups is not expected to dramatically
change key bond lengths in the optimized geometries and relative energies of individual spin
states (Neu, M.; Quesne, G. et al 2014, Lanucara, F.; Chiavarino, B. et al 2011). Thus, recent
work of the Goldberg group showed that meso-substituted manganese-oxo porphyrinoid
complexes retained the spin state ordering and converged to a closed-shell singlet
manganese(V)-oxo state in all cases (Neu, M.; Quesne, G. et al 2014).
116
Figure 4.4: Optimized geometries of the 4,2
A2u and 4,2
A1u states of 4,2
A as calculated
with UB3LYP/BS1 [UB3LYP/BS2] {UB3LYP-D3/BS2} with Fe–O bond lengths in
angstroms.
The calculations on the low-lying 4,2
A2u and 4,2
A1u states reported in Table 4.2 and Fig 4.4
show that geometrically there are very little differences between these states, but the spin
state ordering and relative energies are sensitive to the method and basis set. Recent,
complete active site (CASSCF) and restricted active site (RASSCF) calculations of Pierloot
and co-workers (Radoń, M.; Broclawic, E. et al 2011) calculated the 4,2
A2u and 4,2
A1u states of
A within 1 kcal mol–1
of each other with a small preference for the A1u states. However, they
also located two low-lying iron(V) states, which we were unable to characterize and for
which no experimental evidence exist. Unfortunately, our chemical systems (in particular
structure B) are too large to attempt calculations using the CASSCF and RASSCF methods;
therefore, we decided to continue with UB3LYP instead.Subsequently, we investigated the
substrate epoxidation by [FeIV
(O)(Por+
)]+, i.e.
4A. We find the lowest lying barriers to
proceed from the 4A2u state and will focus on those in the following. Previous studies on the
epoxidation of olefins by [FeIV
(O)(Por+
)(L)] with L = NCCH3 or Cl– showed that the same
2A1u: 1.630 [1.608] {1.608}4A1u: 1.637 [1.611] {1.608}2A2u: 1.626 [1.604] {1.604}4A2u: 1.627 [1.604] {1.604}
117
trends in reactivity are observed when the Por ligand is replaced by TPFPP (D. Kumar, L.
Tahsini, et al 2009, Kumar, D.; Latifi, R. et al 2013), hence the smaller model was used in
this study. We investigated substrate epoxidation with a range of olefins: ethene (1), propene
(2), 1-butene (3), E-2-butene (4), cyclohexene (5), 1,3-cyclohexadiene (6), styrene (7), trans-
-methylstyrene (8), Z-2-butene (9) and 2-pinene (10). For all substrates we calculated the
full potential energy profile from reactants to epoxide products, but for space restrictions we
will focus on the rate determining C–O bond formation transition states (TSCO) only. All
reactions are concerted with a single C–O activation barrier leading to epoxide product
complexes PE. This is unusual as previous calculations on substrate epoxidation by Cpd I
models gave a stepwise mechanism via a radical intermediate that via a ring-closure barrier
was separated from epoxide product complexes (de Visser, P.; Ogliaro, F. et al 2001, de
Visser, P.; Ogliaro, F. et al 2001, de Visser, P.; Ogliaro, F. 2002, Kumar, D.; Karamzadeh, B.
et al 2010, Kumar, D.; de Visser, P. et al 2005). The orientation of the substrate and the
strong displacement of the metal from the porphyrin plane are the likely reason for the fact
that radical intermediates are saddlepoints here. Thus, the ring-closure barrier on the quartet
spin state surface involves an electron transfer from substrate into *z2. The latter orbital in
iron-porphyrin complexes with axial ligand, e.g. thiolate, contains a strong contribution from
axial ligand orbitals (3pz) and therefore is high in energy. Since our particular system lacks
an axial ligand, the *z2 orbital is considerably lower in energy and as a consequence the
lifetime of the radical intermediate is reduced and the reaction to form products is now
concerted.
Fig 4.5 gives the optimized geometries of the C–O activation transition states (TSCO) for
all substrates. Generally, the transition states occur early with a long C–O distance and
118
relatively short Fe–O distance that has not dramatically changed from what it was in the
iron(IV)-oxo porphyrin cation radical state. As expected the metal is considerably displaced
from the plane through the four nitrogen atoms of the porphyrin ring by as much as 0.287 –
0.299 Å. These transition states bear resemblance to substrate epoxidation barriers calculated
previously for P450 Cpd I reactions with olefins (de Visser, S.P.; Ogliaro, F. et al 2001, de
Visser, S.P.; Ogliaro, F. et al 2001, de Visser, S.P.; Ogliaro, F. 2002, Kumar, D.;
Karamzadeh, B. et al 2010, Kumar, D.; de Visser, S.P. et al 2005). Electronically, all
transition states are accomplished by single electron transfer from the substrate into the a2u
orbital and the formation of an [FeIV
(OSub)(TPFPP)]+ transition state with orbital occupation
[core] x2–y22 *xz
1 *yz
1 a1u
2 a2u
2 Sub
1 with Sub a radical on the substrate group.
Figure 4.5: UB3LYP/BS1 optimized geometries of epoxidation transition states with
bond lengths in angstroms.
1.687
1.908
1.390
4TS(1)
i435.9 cm–1
rFeN,average = 2.011average = 0.287
1.7242.041
1.414
4TS(4)
i242.4 cm–1
rFeN,average = 2.003average = 0.292
1.644
2.1931.379
4TS(5)
i91.5 cm–1
rFeN,average = 2.015average = 0.288
1.6492.187
1.380
4TS(7)
i167.5 cm–1
rFeN,average = 2.013average = 0.295
1.663
2.171 1.404
4TS(6)
i214.7 cm–1
rFeN,average = 2.006average = 0.290
1.648
2.237
1.391
4TS(8)
i134.6 cm–1
rFeN,average = 2.012average = 0.299
1.6662.072
1.380
4TS(2)
i303.5 cm–1
rFeN,average = 2.014average = 0.296
1.665
2.073
1.381
4TS(3)
i299.9 cm–1
rFeN,average = 2.015average = 0.298
1.6542.145
1.387
4TS(9)
i248.3 cm–1
rFeN,average = 2.002average = 0.268
1.642
2.3231.380
4TS(10)
i59.2 cm–1
rFeN,average = 2.015average = 0.293
119
4.5 Discussion
The present work gives a detailed and extensive overview on the reactivity of iron(IV)-oxo
porphyrin systems with a test-set of olefins. Crestoni and co-workers determined rate
constants and measured product ion distributions in the gas phase using FT-ICR MS. This
comprehensive set of transition metal containing reactivities is unique and will enable
computation to benchmark and calibrate its methods effectively. This is particularly
important for transition metal complexes, such as iron(IV)-oxo species, where the
reproducibility of the computational (DFT) methods sometimes varies strongly depending on
the density functional method used, the basis set, environmental perturbations, dispersion
effects etc (A. Ghosh and P. R. Taylor, 2003, M. Swart, A. R. Groenhof, et al 2004, de
Visser, P.; Quesne, G. et al 2004). In this work, we supplemented the experimental studies
with a series of preliminary DFT calculations for two reasons: (i) to validate and calibrate
computational methods against experiment; (ii) to establish the physicochemical properties
that influence the rate constant of the chemical reaction.
Let us first start with a comparison of the experimental and computational reaction rates.
As FT-ICR MS experiments are being performed at very low pressures, these experimental
conditions are close to Ideal Gas conditions with very few molecular collisions per second. In
the kinetic study of ion-molecule reactions in the gas phase one needs to consider that
thermal equilibration of the reacting system with the environment is in general not granted.
On the contrary, when small species react at low pressures, the absence of thermalizing
collisions leads to non-equilibrium energy distributions. In the absence of solvation, the
double-well potential model first proposed by Brauman in 1977 to account for the kinetic
behaviour of displacement reactions by anionic nucleophiles predicts that the energy of the
120
intermediate and transition state must lie below the energy of the combined reactants
(Chabinyc, L.; Craig, L. et al 1998). Because at low pressure the intermediates cannot be
stabilized by unreactive collisions, determining the transition state energy is less
straightforward than in solution. However, the kinetics results presently reported deal with a
relatively large reactant ion that effectively establishes thermal equilibrium with the
environment through coupling with the background radiation field allowed by the several low
frequency IR modes of the iron(IV)-oxo macrocyclic ligand complex. This condition is
responsible, for example, for the consistency between the kinetics of NO ligand addition
kinetics to iron(II/III) porphyrin complexes and the equilibrium data independently
established through equilibrium measurements (Angelelli, F.; Chiavarino, B. et al 2005).
Because of these considerations, the notion can be adopted that the presently investigated
systems are in prevailing thermal equilibrium with the environment and reaction kinetics can
be interpreted within the framework of transition state theory. Consequently, reaction rates
represent bimolecular reactions and as such they should compare to computationally
determined reaction rates well.
121
Figure 4.6: Correlation between experimental and computational barrier heights.
0
5
10
15
20
0 5 10 15 20RT ln kexp [kcal mol‒1]
Hcalc
[kcal mol‒1]
122
The experimental rate constants were converted into free energy units via RT ln kexp, with R
the gas constant and T the temperature, using transition state theory and plotted against the
calculated enthalpy of activation for the same substrates, see Fig 4.6. Although only a limited
computational study is reported here, when we calculate the deviation between experiment
and theory for each data point, we find an average difference between experiment and theory
of 1.5 kcal mol–1
with a standard deviation of 3.4 kcal mol–1
. As such, the DFT methods used
here reproduce the trends obtained from experimental enthalpies of activation well and the
linearity and reproducibility of the calculations is well within the typical error reported for
DFT calculations using this method of about 5 kcal mol–1
(Y. Zhao and D. G. Truhlar, et al
2008, Schwabe, T.; Grimme, S. 2008, Weymuth, T.; Couzijn, A. et al 2014). There is,
however, a large systematic error as well as a relatively large standard deviation that require
further studies. Note that the experimental data in Figure 4.6 refers to free energies of
activation, whereas the computational results are enthalpy changes instead. The systematic
error between experiment and theory contains entropic corrections to the energy.
Figure 4.7: (a) Correlation between experimentally determined RT ln kexp (for raw data,
see Table 4.1) versus known ionization energies (IE). (b) Correlation between
calculated epoxidation activation enthalpy (in kcal mol–1
) and experimental ionization
energy for the substrates in Fig 4.5.
y = 0.15x + 7.89
R² = 0.82
6
8
10
12
0 5 10 15 20
IE
[eV]
Hcalc [kcal mol‒1]
(b)
y = -0.382x + 8.55
R² = 0.96
6
8
10
12
-6 -4 -2 0 2
IE
[eV]
RT ln kexp [kcal mol‒1]
(a)
123
Subsequently, we investigated the origin of the rate constant, and in particular, the physical
and chemical properties of the substrate and oxidant that determine the reaction mechanism
and the enthalpy of activation of an epoxidation reaction. Previous studies on heteroatom
oxidation and double bond epoxidation by P450 enzymes implicated a correlation between
the natural logarithm of the rate constant with the ionization energy of the substrate (Crestoni,
E.; Fornarini, S. et al 2009, Lanucara, F.; Crestoni, E. 2011, de Visser, S.P.; Ogliaro, F. et al
2002, Watanabe, Y.; Iyanagi, T. et al 1980). To find out whether the data in Table 4.1 follow
these trends as well, we plot RT ln kexp versus experimentally known ionization energies
(Lias, G.; Bartmess, E. et al), see Fig 4.7. The set of data shown in Table 4.1 and Fig 4.7
gives a linear correlation between the natural logarithm of the rate constant and the ionization
energy of the substrate with an R2 = 0.96. Fig 4.7(b) displays the correlation between the
DFT calculated enthalpy of activation of the reaction of [FeIV
(O)(Por+•
)]+ with olefins. In
agreement with the experimental trends given in part (a) of Fig 4.7 also the computational
trends link the natural logarithm of the rate constant to the ionization energy of the substrate.
Clearly, the key physicochemical property that drives the reaction mechanism and affects the
rate constant of substrate epoxidation by iron(IV)-oxo complexes is the ionization energy of
the substrate.
In order to explain the experimental and computational trends in the reaction mechanisms,
we devised a valence bond (VB) curve crossing diagram, which is schematically depicted in
Fig 4.8. This diagram starts bottom left with the reactant configuration of
[FeIV
(O)(TPFPP+•
)]+ in electronic configuration xz
2 yz
2 x2–y2
2 *xz
1 *yz
1 a2u
1. The and *
electrons along the FeO bond are identified with dots in the VB diagram and due to
occupation of xz2 *xz
1 there are three dots on the left-hand-side of the Fe–O bond. In
addition, there are three electrons in the yz and *yz orbitals, which are identified with the
other three dots on the right-hand-side of the Fe–O bond. Furthermore, the oxidant has a
124
radical on the porphyrin ring for single occupation of the a2u molecular orbital. The substrate
double bond is also highlighted with four electrons spread out over the interaction. Upon
approach of the substrate on the iron(IV)-oxo species a radical intermediate is formed that has
a single bond between the oxygen and carbon atoms and a doubly occupied a2u orbital.
Figure 4.8: VB curve crossing diagram for the C–O bond formation step in olefin
epoxidation (R2C=CH2) by [FeIV
(O)(TPFPP+•
)]+. Valence electrons are identified with
a dot and lines (curved and straight) in the VB structures represent bonds.
In VB theory the electronic configuration in the reactant complex (R) connects to an excited
state in the product geometry (P*) as shown with the blue line in Fig 4.8. At the same time
the product electronic configuration (P) connects to an excited state in the reactant geometry
(R*), black line in Fig 4.8. These two VB curves cross and lead to an avoided crossing and a
transition state for the C–O bond formation with barrier E‡. The barrier height is linearly
proportional to the curve crossing energy, which in its own right is a fraction of the excitation
energy (GH) from the reactant wave function to the product wave function in the geometry of
125
the reactants, i.e. for R R*. The difference in VB structures for R and R* thereby
should give a reflection of the key electron transfer/migrations upon product formation.
Moreover, based on the excitation energy, the factors that determine the barrier height can be
predicted.
An analysis of the differences between the reactant and product wave functions in the
reactant geometry reveals the following information: First of all, a comparison of the VB
structures of R and R* shows that the electrons in the -bond of the olefin are singlet
paired in the ground state and triplet coupled in the excited state, hence the excitation energy
GH includes the -* electron excitation in the substrate, Eex,Sub. Generally, the first
ionization potential of an olefin corresponds to the removal of an electron from a -orbital,
and, hence, is proportional to the * excitation energy. Indeed, our experimentally and
computationally determined barrier heights correlate linearly with the ionization energy of
the olefin, and therefore support the VB model.
One of the electrons originating from the -bond of the olefin forms a bond with the *xz
electron along the FeO bond, to create the C–O bonding pair of electrons. This means that the
xz/*xz pair of orbitals during the reaction splits back into individual atomic orbitals namely
3dxz(Fe) and 2px(O). The 2px(O) electron pairs with the electron from the substrate, while one
of the electrons of the 3dxz(Fe) orbital is transferred into the a2u orbital through internal
excitation/rehybridization of the oxidant, Eex,ox. The promotion gap, GH, therefore, will be
proportional to the * excitation in the substrate and the 3dxz to a2u electron transfer in the
oxidant: GH = Eex,Sub + Eex,ox. Obviously, since the ionization energy represents the energy to
remove an electron from a -type orbital of an olefin, this will imply a linear correlation
between the first excited state and the ionization energy of the substrate (de Visser, S.P.;
Ogliaro, F. et al 2002). The VB diagram, therefore, confirms a linear correlation between the
126
ionization energy of the substrate and the C–O bond formation enthalpy of activation as
shown above.
Although, the oxygen atom transfer reaction between [FeIV
(O)(TPFPP+
)]+ and an olefin
could lead to either epoxide or hydroxylated products, unfortunately the FT-ICR MS
experiments cannot distinguish the two. Thus, several substrates Scheme 4.2 and Table 4.1
contain aliphatic groups that in a reaction with an iron(IV)-oxo group can be converted into
an alcohol. A correlation between the rate constant of oxygen atom transfer and the ionization
energy, Fig 4.7a, of the olefin provides indirect experimental evidence that all reactions lead
to epoxidation products. In fact, hydrogen atom abstraction reactions should not correlate
with the ionization potential of the substrate, but were shown to be proportional to the
strength of the C–H bond of the substrate that is formed (Lanucara, F.; Crestoni, E. 2011,
Watanabe, Y.; Iyanagi, T. et al 1980). To test that the rate constants do not correlate with the
bond dissociation energy (BDECH) of the C–H bond of the substrate that is broken, we plot in
Fig 4.9 calculated BDECH and barrier heights of selected olefins, namely propene, Z-2-
butene, E-2-butene, cyclohexene and 1,3-cyclohexadiene. As can be seen from Fig 4.9, no
correlation between BDECH and barrier height exists, and, therefore, hydrogen atom
abstraction is not the rate determining step in the reaction mechanism. Further evidence that
hydrogen atom abstraction reactions can be ruled out here comes from kinetic isotope effect
(KIE) studies. We measured the rate constant of oxygen atom transfer with cyclohexene and
cyclohexene-d10 and determined a KIE = kH/kD ~ 1 (Table 4.1). Consequently, the oxygen
atom transfer is unlikely to proceed with an initial hydrogen atom abstraction and double
bond epoxidation will be the dominant pathway.
The experimental trends, therefore, provide the first indirect experimental evidence that in
the gas-phase the regioselectivity of double bond epoxidation versus aliphatic hydroxylation
will be in favour of the epoxidation pathway. This implies that in enzymatic systems, such as
127
the cytochromes P450, the shape and size of the substrate binding pocket will influence the
regioselectivity of hydroxylation over epoxidation and can change the natural preference
away from epoxidation.
Finally, the calculations presented in this work obviously refer to gas-phase results and
hence correlate well with gas-phase mass spectrometric data. In order to further establish that
the work can be extrapolated to solution phase, we did a series of single point calculations
using the polarized continuum model with a dielectric constant of = 4.7 to mimic a solution.
The obtained correlation between solvent corrected free energies of activation of epoxidation
reactions by [FeIV
(O)(Por+
)]+ is plotted against the solvent corrected ionization energy of all
substrates although it is not shown in here. Even in solvent, the linear trend in the correlation
between free energy of activation and ionization energy is retained, therefore, we expect to be
able to extrapolate our results to the solution phase as well.
Figure 4.9: Correlation between calculated epoxidation activation enthalpy (in kcal
mol–1
) and BDECH for the substrates.
y = -0.52x + 49.12
R² = 0.34
0
5
10
15
20
60 65 70 75 80 85 90
BDECH [kcal mol–1]
Hcalc
[kcal mol‒1]
128
4.6 Conclusion
In this work we report a comprehensive combined mass spectrometric and computational
study on substrate epoxidation by iron(IV)-oxo porphyrin complexes in the gas phase. We
present a novel method to synthesize [FeIV
(O)(TPFPP+
)]+ in the gas phase at low pressure.
Furthermore, we report a large set of experimentally derived rate constants and product
distributions. All olefins undergo oxygen atom transfer, whereas compounds with low
ionization energy also give a certain degree of hydride transfer and charge transfer reactions.
Our experimentally determined reaction rates correlate linearly with the ionization potential
of the substrate and show that the electron transfer from substrate to oxidant is rate
determining. A thorough computational survey has confirmed the suggested mechanism and
provides a rationale for the observed trend in the rate constants.
129
CHAPTER 5
PROJECT THREE
130
Rationalization of the barrier height for para-Z-
styrene epoxidation by iron(IV)-oxo porphyrins
with variable axial ligands.3
ABSTRACT
A versatile class of heme monoxygenases involved in many vital functions for human health
are the cytochromes P450, which react via a high-valent iron(IV)-oxo heme cation radical
species called Compound I. One of the key reactions catalyzed by these enzymes is C=C
epoxidation of substrates. We report here a systematic study into the intrinsic chemical
properties of substrate and oxidant that affect reactivity patterns. To this end, we investigated
the effect of styrene and para-substituted styrene epoxidation by Compound I models with
either an anionic (chloride) or neutral (acetonitrile) axial ligand. We show, for the first time,
that the activation enthalpy of the reaction is determined by the ionization potential of the
substrate, the electron affinity of the oxidant as well as by the strength of the newly formed
C–O bond (approximated by the bond dissociation energy, BDEOH). We have set up a new
valence bond model that enables us to generalize substrate epoxidation reactions by iron(IV)-
oxo porphyrin cation radical oxidants and make predictions of rate constants and reactivities.
We show here that electron withdrawing substituents lead to early transition states, whereas
electron donating groups on the olefin substrate give late transition states. This affects the
barrier heights in such a way that electron withdrawing substituents correlate the barrier 3 Devesh Kumar, Reza Latifi, Suresh Kumar, Elena V. Rybak-Akimova, Mala A. Sainna, and Sam P. de Visser, “Rationalization of the barrier
height for para-Z-styrene epoxidation by iron(IV)-oxo porphyrins with variable axial ligands” Inorganic Chemistry, 2013, 52 (14), 7968–
7979
131
height with BDEOH, while the electron affinity of the oxidant is proportional to the barrier
height for substrates with electron donating substituents.
5.1 Introduction.
An important class of enzymes in human physiology are the cytochromes P450, which are
a large set of heme enzymes involved in the biodegradation and metabolism of toxic
compounds in the liver (Sono, M.; Roach, M. et al 1996, Groves, J. 2003, Ortiz de
Montellano, P. 2004, Kadish, K.; Smith, K. et al 2010, Ortiz de Montellano, P. 2010, de
Visser, S.; Kumar, D. 2011). These enzymes utilize molecular oxygen on a heme center in a
catalytic cycle that uses two electrons and two protons to generate an iron(IV)-oxo heme
cation radical active species called Compound I (CpdI) (Denisov, I.; Makris, T. et al 2005,
Rittle, J.; Green, M. 2010). This species is the active oxidant of the P450 enzymes and reacts
with substrates via, for instance, aliphatic and aromatic hydroxylation, double bond
epoxidation, N-dealkylation and sulfoxidation (Groves, J.; Shalyaev, K. et al 2000, Nam, W.
2007). Because of its versatility in substrate activation the group of enzymes has attracted
interest from the biotechnological and pharmaceutical industries although the intricate details
of its catalytic mechanism and reactivity with substrates are still poorly understood. The
epoxidation of olefins is a common reaction in P450 enzymes for a variety of important
bioprocesses in the body including the activation of unsaturated fatty acids (Ruettinger, R.;
Fulco, A. 1981, Guengerich, F. 2003, McLean, K.; Munro, A. 2008). As a result, substrate
epoxidation by P450 isozymes is well studied for a wide range of (non)natural substrates
(Groves, J.; Avaria-Neisser, G. et al 1986, Alcalde, M.; Farinas, E. et al 2004, Dansette, P.;
Bertho, G. et al 2005, Yuan, X.; Wang, Q. et al 2009, Shinkyo, R.; Xu, L. et al 2011). For
instance, the reactions were shown to be highly enantioselective, whereby cis--methyl
styrene substrate gave an 89:11 epoxide product ratio of the 1S,2R over the 1R,2S form (Ortiz
de Montellano, P.; Fruetel, J. et al 1991).
132
Since CpdI is a very versatile oxidant, many studies investigated synthetic analogues and
models in substrate oxidation (Groves, J.; Myers, R. et al 1983, Groves, J.; Watanabe, Y.
1986, Collman, J.; Kodadek, T. et al 1986, Ostović, D.; Bruice, T. 1989, Collman, J.;
Brauman, J. et al 1990, Groves, J.; Gross, Z. et al 1994, Stephenson, N.; Bell, A. 2006,
Collman, J.; Zeng, L. et al 2006, Hessenauer-Ilicheva, N.; Franke, A. et al 2007, Mas-
Balleste, R.; Que Jr., L. 2007, Comba, P.; Rajaraman, G. 2008, Bruijnincx, P.; Buurmans, I.
et al 2008, Hull, J.; Sauer, E. et al 2009, McGown, A.; Kerber, W. et al 2009, Company, A.;
Feng, Y. et al 2009, Franke, A.; Wolak, M. et al 2009, Hessenauer-Ilicheva, N.; Franke, A. et
al 2009, Leeladee, P.; Goldberg, D. 2010, Lanucara, F.; Crestoni, M. 2011). A number of
these biomimetic studies focused on double bond epoxidation mechanisms and one of the
most common substrates in these studies is styrene. Thus, a series of styrene epoxidation
studies with a CpdI mimic with varying axial ligand gave rate constants that were
proportional to its electron donating ability (Gross, Z.; Nimri, S. 1994, Gross, Z. 1996,
Czarnecki, K.; Nimri, S. et al 1996). Green and co-workers (Green, M.; Dawson, J. et al
2004) characterized the axial ligand effect as arising from changes in pKa values of the oxo
group, which was confirmed with density functional theory (DFT) calculations on model
complexes (Kumar, D.; Karamzadeh, B. et al 2010).
Experimental studies reported rate constants for styrene epoxidation by synthetic iron(IV)-
oxo complexes: [FeIV
(O)(TPFPP+•
)X]0/+
with TPFPP = meso-
tetrakis(pentafluorophenyl)porphyrinato and X = Cl– or NCCH3 (Song, W.; Ryu, Y. et al
2005). In an intriguing set of experiments, it was shown that the reaction of
[FeIV
(O)(TPFPP+•
)NCCH3]+ with ethylbenzene gave aromatic hydroxylation products,
whereas the one using [FeIV
(O)(TPFPP+•
)Cl] as an oxidant led to benzylic hydroxylation
instead, hence the axial ligand in iron(IV)-oxo porphyrins seems to affect the product
distributions in the reaction processes and the regioselectivity of aromatic over aliphatic
133
hydroxylation (Song, W.; Ryu, Y. et al 2005). The two oxidants with chloride versus
acetonitrile as axial ligand also gave differences in styrene epoxidation using para-substituted
styrene derivatives. The work identified a correlation between the rate constant of styrene
epoxidation with the + Hammett factor, but the slopes were different for the CpdI models
with anionic versus neutral axial ligands. The fundamental nature of this axial ligand effect
and how it affects reaction mechanisms, rate constants and product distributions of iron(IV)-
oxo porphyrins is unknown and, therefore, warrant a computational study. Thus, to gain
insight into the effect of axial ligands on the reactivity of iron(IV)-oxo porphyrins with
olefins we have done a density functional theory study of the activation barriers using a
selection of para-substituted styrene as substrates and Compound I models with meso-
substituted porphyrin.
Fujii and co-workers did a systematic investigation into the redox potentials of
[FeIV
(O)(TMP)X]n+
with TMP = 5,10,15,20-tetramesitylporphyrinate and X an axial ligand
that is either anionic or neutral (Takahashi, A.; Kurahashi, T. et al 2011, Cong, Z.; Kurahashi,
T. et al 2011). They found a positive E½ shift upon binding of an anionic axial ligand that was
virtually constant for a range of ligands. By contrast, binding of neutral ligands, such as
imidazole gave redox potentials in line with those found for peroxidases corresponding to a
negative E½ shift. They then replaced the TMP group with a meso-substituted porphyrin
ligand with electron-withdrawing groups, which led to an increase of the E½ values.
Currently, it is not clear what factors determine the redox potentials of iron-porphyrins and
whether there is a relationship with catalysis. Therefore, to gain insight into the effect of
meso-substitution on the intrinsic chemical properties of the oxidant and the subsequent
catalysis of substrates we decided to include this study in the present work.
The model we chose for our studies presented here is given in Figure 5.1 and is based on
the system used in Ref (Song, W.; Ryu, Y. et al 2005). The basic features of the oxidant are
134
an iron(IV)-oxo group embedded in a protoporphyrin IX (Por) without side chains and with
either chloride or acetonitrile as the axial ligand: [FeIV
(O)(Por+•
)X]0/+
with X = Cl– or
NCCH3. In addition, we also investigated a more elaborate model that uses pentafluorophenyl
substituents on the meso-position of the porphyrin ring, i.e. meso-
tetrakis(pentafluorophenyl)porphyrinato or TPFPP. Substrate epoxidation was studied using a
range of para-Z-substituted styrenes with Z = H, F, Cl, CH3, t-Bu, CN, NO2, OCH3, NH2 and
N(CH3)2. We show that para-substitution affects the ionization potential of the substrate and
its electron donating ability, which leads to changes in the activation barrier of oxygen atom
transfer. Although we reported a detailed analysis of substrate epoxidation by iron(IV)-oxo
porphyrins before (Kumar, D.; Karamzadeh, B. et al 2010), the systematic study described
here goes beyond that of the previous work and highlights the differences in reactivity of
substrates with electron donating versus electron withdrawing substituents. Moreover, a new
model is presented that correlates with the obtained reactivity trends.
Figure 5.1: Oxidants and substrates used in this work.
5.2 Methods
The study presented here uses density functional theory methods as implemented in the
Jaguar, Gaussian-03 and Gaussian-09 program packages (Schrödinger, 2011). All geometries
are the result of a full geometry optimization, whereby all degrees of freedom are minimized.
Z
Substrates:Z = H, F, Cl, CH3, t-Bu, CN,NO2, OCH3, NH2, N(CH3)2
FF
F
FF
Y = H,
NH N
HNN
Y
Y
Y
Y
Por: Y = H; TPFPP: Y =
FF
F
FF
X = Cl- or NCCH3
X
135
A subsequent analytical frequency calculation characterized the structures as local minima
(with real frequencies only) or first order saddle points with one imaginary frequency for the
correct mode. The hybrid B3LYP method (Becke, A. 1993, Lee, C.; Yang, W. et al 1988) was
employed throughout in combination with the Los Alamos-type LACVP basis set on iron and
6-31G on the rest of the atoms (BS1) for the geometry optimizations and frequencies (Hay, P.
J.; Wadt, W. 1985, Hehre, W. J.; Ditchfield, R. et al 1972). Energies were then improved by
single point calculations with a triple- type basis set on iron (LACV3P+) and 6-311+G* on
the rest of the atoms; BS2. Benchmark studies against experimental data reproduced free
energies of activation using these methods to within 3 kcal mol–1
(de Visser, S.; Oh, K. et al
2007, Vardhaman, A.; Sastri, C. et al 2011). Previously, we calculated a full potential energy
profile for substrate hydroxylation by an iron(IV)-oxo porphyrin cation radical system at
UB3LYP/BS2 and obtained relative energies within a few tenth of a kcal mol–1
for local
minima and first order saddle points along a reaction mechanism as compared to those
obtained at UB3LYP/BS2//UB3LYP/BS1, hence the latter method was used here (de Visser,
S. 2010). Single point calculations in a dielectric constant with = 37.5 mimicking
acetonitrile with a probe radius of 2.1 Å were performed in Jaguar using basis set BS2. Free
energies reported here were calculated at 298K temperature and 1 bar pressure and are based
on the UB3LYP/BS2//UB3LYP/BS1 energies and corrected with ZPE, thermal and entropic
corrections from the frequency file and with solvent corrections from the single point solvent
calculation. Further corrections to the energy were made by performing single point
calculations with dispersion corrected B3LYP as implemented in Jaguar (Schwabe, T.;
Grimme, S. 2007). For selected structures we also did geometry optimizations and
frequencies using UB3LYP-D/BS1 in Jaguar, although little changes in the optimized
geometries were obtained.
136
We used two synthetic iron-porphyrin models that are distinguished by the choice of the
axial ligand, which was either chloride or acetonitrile. Note that [FeIV
(O)(Por+•
)Cl]0 is overall
charge neutral, while [FeIV
(O)(Por+•
)NCCH3]+ is positively charged. In our initial
calculations we used a bare porphyrin ring, whereby all side chains were replaced by
hydrogen atoms. In a second set of calculations we studied a TPFPP ligand system, TPFPP =
tetra(pentafluorophenyl)porphyrin. We studied styrene epoxidation using a selection of para-
substituted styrene substrates as described in Figure 5.1.
To ascertain that the results are not influenced by the density functional method chosen
here, we ran a selection of single point calculations using dispersion corrected DFT
(Schwabe, T.; Grimme, S. 2007) and B3LYP with 15% HF exchange (designated B3LYP*)
(Reiher, M.; Salomon, O. et al 2001). As before (Kumar, D.; Thiel, W. et al 2011), these test
calculations reproduced the trends obtained with the B3LYP method so that the results give a
systematic error, which does not affect our discussion and analysis as we are dealing with
trends only here. Nevertheless, for comparison we give here the results obtained for E+ZPE
(UB3LYP), G + Esolv (UB3LYP) and G + Esolv + Edisp (UB3LYP-D).
5.3 Results
We started the work with a detailed study into the electronic properties of iron(IV)-oxo
complexes, the effects of axial versus equatorial ligands and finally the reactivity patterns
with a selection of para-substituted styrenes. Our models include [FeIV
(O)(Por+•
)X]0/+
,
designated 1X, and [FeIV
(O)(TPFPP+•
)X]0/+
, designated 2X, with X = Cl– and NCCH3, Figure
5.1. Similar to previous studies on CpdI of P450 and biomimetic iron(IV)-oxo porphyrin
complexes (Green, M. 1999, Ogliaro, F.; de Visser, S. et al 2001, Kamachi, T.; Yoshizawa,
K. 2003, Bathelt, C.; Zurek, J. et al 2005, Lonsdale, R.; Oláh, J. et al 2011, Isobe, H.;
Yamanaka, S. et al 2011), all complexes have the same electronic ground state with four
137
electrons in metal 3d-type orbitals and a radical on the porphyrin macrocycle. High lying
occupied molecular orbitals include the x2–y2 orbital that is nonbonding, doubly occupied and
located in the plane of the porphyrin. Slightly higher in energy are two *FeO orbitals (*xz,
*yz) for the antibonding interaction of the metal 3dxz/yz with 2px/y atomic orbitals on oxygen.
Higher lying and virtual are the *z2 and *xy orbitals for the antibonding interactions along
the Fe–O axis and between the Fe–N groups in the plane of the porphyrin ring. In addition
there is a radical on a porphyrin type orbital that in D4h symmetry has the label a2u. The
orbital occupation gives close lying doublet and quartet spin states with configuration x2–y22
*xz1 *yz
1 a2u
1, whereby the two * electrons are either ferromagnetically or
antiferromagnetically coupled to the a2u electron in the quartet and doublet spin states. As a
consequence the two spin states are close in energy and the oxidant reacts via two-state-
reactivity patterns on close-lying doublet and quartet spin state surfaces (Shaik, S.; Kumar, D.
et al 2005).
138
Figure 5.2: UB3LYP/BS1 optimized geometries of 4,2
1X and 4,2
2X in the gas phase with bond
lengths in angstroms. Group spin densities are obtained at UB3LYP/BS2//UB3LYP/BS1 and
are reported in atomic units.
Figure 5.2 gives optimized geometries of 4,2
1X and 4,2
2X with X = Cl–/NCCH3 as calculated
with DFT. Optimized geometries of 4,2
1X (X = Cl–/NCCH3) are almost identical to those
reported before (de Visser, S.; Tahsini, L. et al 2009). The Fe–O distances are short and
typical for an iron(IV)-oxo species and shorten somewhat with an axially ligated acetonitrile
molecule as compared to chloride. This is due to mixing of the a2u and * orbitals in 4,2
1Cl
41Cl (21Cl)
42Cl (22Cl)
41NCCH3 (21NCCH3)
42NCCH3 (22NCCH3)
rFeO = 1.660 (1.657)
rFeCl = 2.419 (2.433)
rFeO = 1.660 (1.658)
rFeCl = 2.424 (2.436)
rFeO = 1.646 (1.645)
rFeNax = 2.126 (2.129)
Fe = 1.00 (1.12)
O = 1.02 (0.96)
Cl = 0.19 (–0.17)
TPFPP = 0.80 (–0.92)
Fe = 1.03 (1.14)
O = 0.99 (0.95)
Cl = 0.15 (–0.13)
Por = 0.83 (–0.95)
Fe = 0.97 (1.01)
O = 1.05 (1.05)
NCCH3 = 0.00 (–0.02)
Por = 0.97 (–1.04)
rFeO = 1.648 (1.647)
rFeNax = 2.112 (2.121)
Fe = 0.95 (0.99)
O = 1.07 (1.07)
NCCH3 = 0.00 (–0.01)
TPFPP = 0.98 (–1.05)
139
with 3p atomic orbitals on the chloride ligand (de Visser, S.; Tahsini, L. et al 2009), which
brings the metal more inside the plane through the porphyrin ring. This type of mixing has
been identified before as the key reason for the intrinsic electronic differences of CpdI in
monoxygenases and peroxidases (de Visser, S.; Shaik, S. et al 2003). We did additional
geometry optimizations of 21X (X = Cl
–/NCCH3) at UB3LYP-D/BS1 level of theory, which
reproduced the UB3LYP/BS1 structures within 0.010 Å.
As can be seen, meso-substitution of the porphyrin ring with pentafluorophenyl groups has
little effect on the optimized geometries and the Fe–O, Fe–Cl and Fe–Nax distances for
structures 1 and 2 are almost the same. Note also that very little saddling is found for the
[FeIV
(O)(TPFPP+•
)X] structures. Not surprisingly, since the same molecular orbitals are
singly occupied in the doublet and quartet spin state structures, their geometries are virtually
identical for each CpdI set of data. To establish whether the meso-pentafluorophenyl groups
give electronic changes to the iron(IV)-oxo species we also show in Figure 5.2 the group spin
densities (), but only minor differences are observed between the data for structures 1 and 2.
Subsequently, we calculated styrene epoxidation using a range of para-substituted styrene
derivatives. Before we discuss the results on the substituted styrenes, let us focus on the
general overall mechanism first. All epoxidation reactions described in this work proceed
with the same stepwise mechanism that starts with the formation of reactant complex
between CpdI and substrate, R. As an example, we give the reaction profile for
[FeIV
(O)(Por+
)Cl] with para-H-styrene in Figure 5.3 on the lowest doublet and quartet spin
states. The spin multiplicity is given in superscript and in subscript we give the axial ligand
(X) and the para-Z-substituent of styrene. The mechanism resembles that found in earlier
studies of substrate epoxidation of olefins by metal(IV)-oxo oxidants (de Visser, S.; Tahsini,
L. et al 2009, Linde, C.; Åkermark, B. et al 1999, Kamachi, T.; Shiota, Y. et al 2003,
Quiñinero, D.; Musaev, D. et al 2003, Sharma, P.; de Visser, S. et al 2003, Hirao, H.; Kumar,
140
D. et al 2005, Kumar, D.; Derat, E. et al 2005, Bassan, A.; Blomberg, M. et al 2005, de
Visser, S.; Ogliaro, F. et al 2001, de Visser, S.; Ogliaro, F. et al 2001, de Visser, S.; Ogliaro,
F. et al 2002, de Visser, S.; Ogliaro, F. et al 2002, Kumar, D.; de Visser, S. et al 2005) and
starts with an initial electrophilic addition of the oxo group to the double olefinic bond of
styrene via a transition state TSX,Z to form a radical intermediate (IX,Z), whereby the subscript
X refers to the axial ligand, i.e. X = Cl– or AN (acetonitrile) and the subscript Z to the para-
substituent of styrene. In a final reaction step a ring-closure transition state (TSrc,Z) leads to
epoxide product complexes (PX,Z). Optimized geometries obtained at UB3LYP-D/BS1 show
little difference with those obtained at UB3LYP/BS1, Figure 5.3.
Figure 5.3: Potential energy profile of styrene epoxidation by 4,2
1Cl as calculated with
UB3LYP/BS2//UB3LYP/BS1. All energies are in kcal mol–1
relative to isolated reactants in
the doublet spin state and include ZPE corrections. Also shown are optimized geometries of
critical points with bond lengths in angstroms and the imaginary frequency in the transition
state in wave numbers. Free energies are given in parenthesis and include UB3LYP/BS2
21Cl+styrene-H
41Cl+styrene-H
2ICl,H
4ICl,H
4TSCl,H
2TSCl,H
0.0 (0.0)
–0.1 (–0.4)
–15.0 (–3.5)
–9.6 (3.8)
7.4 (18.7)
[–1.2 (10.8)]
8.4 (19.4)
4TSrc,H
2TSrc,H
4PCl,H
2PCl,H
2TSCl,H (4TSCl,H)
rFeO = 1.682 (1.706)
[1.674]
rFeCl = 2.387 (2.371) [2.373]
rCO = 2.121 (2.072)
[2.197]
i563.2 (i437.1) [i424.4] cm–1
2ICl,H (4ICl,H)
rFeO = 1.784 (1.804)
rFeCl = 2.322
(2.340)
rCO = 1.472 (1.461)
–15.0
–8.2
–36.6 (–23.7)
–44.6 (–33.6)
CH2 CH
C6H5
C6H5
CH
CH2
C6H5
CHCH2
FeIV
O
X FeIV
O
X
FeIII
O
4,2I4,2P
X
141
energies corrected with thermal and entropic corrections at 298 K. Data in square brackets
obtained after a UB3LYP-D/BS1 geometry optimization.
Although the ring-closure barrier was calculated for a selection of chemical systems,
in all cases its barrier was much smaller than the C–O bond formation barrier via TSX,Z,
therefore, we will focus here on the rate determining barrier only. In particular, in the quartet
spin state ring-closure barriers were located and found to be small, whereas in the doublet
spin state the ring-closure was virtually barrierless. The origin of this difference is due to
differences in electron transfer processes in the ring-closure step, whereby on the doublet spin
state surface the *xz orbital is filled with an extra electron, while on the quartet spin state a
higher lying and virtual *z2 orbital is filled with one electron (de Visser, S.; Ogliaro, F. et al
2001). Energies obtained for the reaction of [FeIV
(O)(Por+•
)Cl] with para-H-styrene are very
similar to those reported before using [FeIV
(O)(Por+•
)SH] as an oxidant (de Visser, S.;
Tahsini, L. et al 2009, de Visser, S.; Ogliaro, F. et al 2001). That is not surprising, since the
pKa and electron affinity of [FeIV
(O)(Por+•
)X], X = Cl–/SH
– are virtually the same.
Geometries are typical for epoxidation structures from previous calculations for P450
catalyzed reaction mechanisms (Linde, C.; Åkermark, B. et al 1999, Kamachi, T.; Shiota, Y.
et al 2003, Quiñinero, D.; Musaev, D. et al 2003, Sharma, P.; de Visser, S. et al 2003, Hirao,
H.; Kumar, D. et al 2005, Kumar, D.; Derat, E. et al 2005, Bassan, A.; Blomberg, M. et al
2005, de Visser, S.; Ogliaro, F. et al 2001, de Visser, S.; Ogliaro, F. et al 2001, de Visser, S.;
Ogliaro, F. et al 2002, de Visser, S.; Ogliaro, F. et al 2002, Kumar, D.; de Visser, S. et al
2005). In the transition state the Fe–O bond elongates slightly with respect to reactants and
further extends in the intermediate to 1.784 Å. At the same time considerable shortening of
the C–O distance occurs to a formally single bond in 4,2
ICl,H of 1.472 Å.
Subsequently, we investigated the reaction mechanisms of [FeIV
(O)(Por+•
)X]0/+
, X = Cl– or
NCCH3 with para-substituted styrene and the optimized geometries of the transition states
142
are given in Figure 5.4. Interestingly, structural differences are found between the two axially
ligated oxidants, whereby the Cl– bound TSX,Z structures are with styrene in an upright
position, whereas the acetonitrile bound ones are more sideways bound. Test calculations
using an upright starting structure and acetonitrile axial ligand, however, converged back to
the sideways bound structure instead, which implicates that the upright structures are higher
in energy for the axially ligated acetonitrile TSs. The differences in substrate orientation
affect the bond distances of the respective transition states. For instance, the bond forming C–
O distances vary from 2.017 – 2.241 Å for the transition states with X = Cl–, whereas for X =
NCCH3 distances between 2.083 and 2.439 Å are found. This implies that the barriers for the
acetonitrile ligated structures are somewhat earlier than the ones for the chloride bound
systems. Although the variation in Fe–O and Fe–X distance is considerably smaller than
those for the C–O distance variations between the oxidant with chloride and acetonitrile
ligands are also found.
Figure 5.4: Optimized geometries of rate determining transition states 2TSX,Z for the reaction
of para-Z-styrene with 21X (X = Cl
–/NCCH3). Geometries optimized at UB3LYP/BS1 with
bond lengths given in angstroms and the imaginary frequency in wave numbers. Also given
Z H F Cl CH3 t-Bu CN NO2 OCH3 NH2 N(CH3)2
X = Cl–
rC–O 2.121 2.107 2.090 2.142 2.143 2.053 2.017 2.163 2.232 2.241
rO–Fe 1.682 1.684 1.685 1.680 1.680 1.689 1.692 1.679 1.674 1.674
rFe–Cl 2.387 2.386 2.385 2.390 2.389 2.383 2.382 2.392 2.399 2.401
imag i563.2 i600.4 i641.0 i500.3 i500.1 i693.3 i697.8 i421.5 i269.4 i245.7
E+ZPE 7.40 7.01 7.06 6.79 6.85 7.30 7.28 5.97 4.80 3.92
G+Esolv+Edisp
13.8 11.2 11.9 9.4 11.3 12.7 11.8 10.1 6.5 –0.3
X = NCCH3
rC–O 2.187 2.178 2.160 2.244 2.272 2.117 2.083 2.305 2.430 2.439
rO–Fe 1.668 1.673 1.677 1.659 1.653 1.692 1.703 1.653 1.648 1.646
rFe–NCCH3 2.103 2.099 2.096 2.115 2.128 2.088 2.084 2.130 2.150 2.163
imag i209.8 i220.3 i234.3 i181.1 i238.2 i274.4 i310.5 i147.8 i75.3 i66.0
E+ZPE 3.55 4.15 4.66 1.39 1.25 8.30 9.44 -1.94 -7.11 -10.56
G+Esolv+Edisp
6.3 8.7 10.7 5.7 7.7 24.8 27.0 4.9 –1.0
Z
Cl
Z
143
are barrier heights (E+ZPE) for 2TSX,Z with energies calculated at
UB3LYP/BS2//UB3LYP/BS1+ZPE relative to isolated reactants in kcal mol–1
and free
energies of activation in solvent (G+Esolv+Edisp) relative to 2RCX,Z.
The epoxidation barriers (TSCl,Z) range from 3.92 kcal mol–1
for para-N(CH3)2-styrene to
7.40 kcal mol–1
for para-H-styrene, i.e. vary by 3.48 kcal mol–1
upon para-substitution. With
acetonitrile as axial ligand the lowest barrier is 10.56 kcal mol–1
below isolated reactants for
para-N(CH3)2-styrene, although it should be noted that the reactant complex is more stable
than isolated reactants by 14.75 kcal mol–1
. Thus, the epoxidation barrier height varies by 20
kcal mol–1
between para-N(CH3)2-styrene and para-NO2-styrene for TSAN,Z, hence the
substituent located at a distance of over 4 Å from the reaction center causes a rate constant
change by a factor of ca. 1015
(estimated using transition state theory for an enthalpy-derived
free energy change of 20 kcal mol–1
; entropy of activation was not included in this estimate).
The ordering of the barrier heights is virtually the same for the two oxidants studied, namely
Z = N(CH3)2 < NH2 < OCH3 < CH3/t-Bu < F/Cl < CN < H. A plot of the barriers 4TSCl,Z
versus those calculated for 4TSAN,Z with the same methods gives a linear correlation.
Furthermore, calculations using a solvent model included give free energies of activation that
follow the same trends as values found in the gas-phase. Inclusion of dispersion corrections
on the solvent corrected free energies of activation gives a further systematic change of the
energetics. This highlights the fact that for the assignment of reactivity trends it does not
matter whether E+ZPE, G, G+Esolv or G+Esolv+Edisp energies are used. As the enthalpies
in a reaction mechanism are determined by the electronic changes of the reactants and we aim
to establish the intrinsic chemical properties that determine the reactivity patterns, we will
focus in the following on establishing trends through the E+ZPE sets of data.
144
To understand the changes in transition state geometry between the para-substituted
styrene substrates, we plot in Figure 5.5 a selection of bond distances versus the barrier
height (E‡+ZPE) for the two CpdI models with either chloride or acetonitrile axial ligand.
Figure 5.5(a) shows the distance of the C–O bond that is formed in the process versus the
barrier height. An almost perfect linear correlation (R2 = 0.98) is obtained for the acetonitrile
ligated TSs and a satisfactory linear correlation is found for the chloride ligated system. Thus,
a drop in C–O bond length of 0.224 Å is found between the weakest epoxidating substrate
(para-NO2-styrene) and the strongest one (para-N(CH3)2-styrene) using a chloride axial
ligand, whereas the difference between these substrates is 0.256 Å for an axially ligated
acetonitrile molecule. This implies that the para-substituent of styrene has a single possibly
electrostatic effect on the C–O bond formation step and consequently the barrier height of the
reaction. To further ascertain that the optimized geometries and barrier heights are
reproducible we reoptimized a selection of TS structures with UB3LYP-D/BS1. Thus,
2TSAN,CN and
2TSCl,CN give Fe–O distances of 1.683 and 1.680 Å and C–O distances of 2.131
and 2.107 Å, respectively. These values are close to those given in Figure 5.4 calculated at
UB3LYP/BS1 level of theory.
y = -14.653x + 37.661
R² = 0.7678
y = -51.372x + 116.46
R² = 0.9767
-15
-10
-5
0
5
10
15
2.0 2.1 2.2 2.3 2.4 2.5
Cl
AN
E‡+ZPE
[kcal mol–1]
C–O distance [angstroms]
(a)
y = 0.0068x + 2.9263
R² = 0.8647
y = 0.0771x - 13.783
R² = 0.9426
-15
-10
-5
0
5
10
15
0 200 400 600 800
Cl
AN
E‡+ZPE
[kcal mol–1]
imaginary frequency[cm–1]
(c)
145
Figure 5.5: Analysis of structural features of the transition states TSX,Z calculated at
UB3LYP/BS1 as a function of the height of the epoxidation barrier with respect to the: (a) C–
O distance, (b) Fe–X distance, and (c) imaginary frequency in the transition state. Data given
for X = Cl– axial ligand (diamonds) and X = acetonitrile (squares).
We also investigated the correlations of epoxidation barrier height with the Fe–O and Fe–X
distances in the transition states, and indeed these correlations are linear as well. With a
chloride axial ligand modest Fe–O and Fe–Cl changes are seen throughout the series of –
0.018 and +0.019 Å, respectively, between the outer ranges for para-NO2-styrene and para-
N(CH3)2-styrene. Much larger differences are found for the weakly bound acetonitrile
system, where the Fe–NCCH3 distance is elongated from 2.084 to 2.163 Å (+0.079 Å shift)
between para-NO2-styrene and para-N(CH3)2-styrene. Thus, the para-substituent of styrene
affects bond distances well over 5 Å from the catalytic center and weakens the metal to axial
ligand distances Fe–Cl and Fe–NCCH3 bonds dramatically.
Interestingly, even the value of the imaginary frequency gives a linear correlation with
barrier height, which implies that the para-substituent of styrene affects the height as well as
the width of the potential energy curve around the transition state. Since, both height and
width of the barrier change linearly for our series of substrates, this means that the area under
the curve could stay the same for these substrates. However, changes to the width of the
y = -176.72x + 428.7
R² = 0.9089
y = -233.96x + 496.28
R² = 0.9647
-15
-10
-5
0
5
10
15
2.0 2.1 2.2 2.3 2.4 2.5
Cl
AN
E‡+ZPE
[kcal mol–1]
Fe–X distance [angstroms]
(b)
146
potential energy curve may have important effects on, for instance, kinetic isotope effects for
a reaction (Kumar, D.; de Visser, S. et al 2003, Kumar, D.; de Visser, S. et al 2004, Kumar,
D.; de Visser, S. et al 2004, de Visser, S. 2006). Thus, a small imaginary frequency correlates
with a broad and wide potential energy surface, whereas a large imaginary frequency
implicates a narrow and high peak. Tunnelling through a narrow and high peak should be
easier than through a broad peak, therefore, kinetic isotope effects associated with this
reaction may be affected as well. It is interesting to note that the imaginary frequencies are
substantially larger for [FeIV
(O)(Por+•
)Cl] than for [FeIV
(O)(Por+•
)NCCH3]+, hence the
barriers for the latter oxidant will be much broader in shape. If this trend also applies to
hydrogen atom abstraction reaction then this would imply that the calculations predict
[FeIV
(O)(Por+•
)Cl] to react with considerably larger kinetic isotope effect for the replacement
of hydrogen by deuterium atoms than [FeIV
(O)(Por+•
)NCCH3]+ with substrates. Indeed, Nam
and co-workers reported differences in KIE values for the two oxidants in aliphatic hydrogen
atom abstraction reactions (Takahashi, A.; Kurahashi, T. et al 2011, Cong, Z.; Kurahashi, T.
et al 2011).
In a final set of calculations, we investigated the effect of meso-substitution of the
porphyrin ring on the electronic properties of CpdI and the epoxidation of para-substituted
styrenes. We selected the TPFPP ligand system with pentafluorophenyl groups on the meso-
position of the porphyrin and calculated styrene epoxidation using the TPFPP oxidant:
[FeIV
(O)(TPFPP+•
)X]0/+
with X = Cl– and NCCH3 or 2X. We studied styrene epoxidation for
para-N(CH3)2-styrene and para-NO2-styrene and compared the mechanism and energy
profiles with those observed above for 1X. Figure 5.6 displays optimized geometries of the
epoxidation transition states TS′X,Z for the reaction of styrene with 2X,Z with Z = NO2 or
N(CH3)2 as calculated with DFT methods.
147
Figure 5.6: Optimized geometries of epoxidation transition states 2TS′X,Z for the reaction of
2X (X = Cl–/NCCH3) with para-Z-styrene. Bond lengths are in angstroms and the value of the
imaginary frequency in wave numbers.
Geometrically, there are striking differences between TSX,Z, on the one hand, with TS′X,Z
on the other hand. Thus, the epoxidation barriers with TPFPP ligand system are characterized
with long C–O and short Fe–O distances, which implicates much earlier transition states
along the potential energy surface. Furthermore, both acetonitrile and chloride ligated TS′X,Z
structures are in upright configuration with a structure similar to the TSCl,Z geometries
reported in Figure 5.4. Probably, the stereochemical interactions with meso-substituents
prevent a substrate orientation analogous to TSAN,Z with a sideways attack on the oxo group.
In addition, the structures are stabilized with hydrogen bonding interactions of substrate C–H
groups with the fluoride atoms from the TPFPP ligand, and specifically those located in the
ortho-position of the meso-substituent. As a consequence, both the reactant complexes as
well as the transition states for epoxidation are considerably stabilized with respect to isolated
reactants. Each transition state structure in Figure 5.6 is stabilized by at least two or three C–
H---F hydrogen bonding interactions with distances between 2.4 – 2.6 Å. F–H hydrogen
bonding interactions can be dramatic and, for instance, trifluoromethanol as a solvent has
2TS′Cl,N(CH3)2 (2TS′AN,N(CH3)2)
rFeO = 1.667 (1.660)
rFeX = 2.394 (2.156)
rCO = 2.574 (2.188)
i53.9 (i162.8) cm–1
2TS′Cl,NO2 (2TS′AN,NO2)
rFeO = 1.676 (1.651)
rFeX = 2.380 (2.118)
rCO = 2.205 (2.305)
i399.3 (i322.8) cm–1
rHF = 2.551 (2.589)
rHF = 2.530 (2.524)rHF = 2.478
rHF = 2.455 (2.470)
E‡+ZPE = –3.8 (–19.8) kcal mol–1E‡+ZPE = 4.5 (7.0) kcal mol–1
148
been shown to lead to considerable lowering of epoxidation and halogenation barriers heights
due to hydrogen bonding and charge transfer interactions (Ben-Daniel, R.; de Visser, S. et al
2003, de Visser, S.; Kaneti, J. et al 2003).
Recent computational studies of dehydrogenation of cyclohexadiene by
[FeIV
(O)(TPCPP+•
)Cl] with TPCPP = meso-tetrakis(pentachlorophenyl)porphyrin also
showed substrate stabilization due to weak hydrogen bonding interactions of chloride atoms
of the TPCPP ligand with C–H groups of the approaching substrate (Kumar, D.; Tahsini, L.
et al 2009). These interactions were shown to be particularly strong for epoxidation barrier
heights due to closer approach to the oxidant as compared to hydrogen atom abstraction
reactions and lowered epoxidation barriers significantly (de Visser, S. 2006). In line with this
it is not surprising that we find considerable lowering of the epoxidation barriers from 7.3 to
4.5 kcal mol–1
for X = Cl/Z = NO2, from 9.4 to 7.0 kcal mol–1
for X = AN/Z = NO2, from 3.9
to –3.8 kcal mol–1
for X = Cl/Z = N(CH3)2 and from –10.6 to –19.8 kcal mol–1
for X = AN/Z
= N(CH3)2 upon replacement of the equatorial ligand from Por to TPFPP. The big changes in
geometry for TSX,Z as compared to TS′X,Z also affect the shape of the potential energy
surface and the values of the imaginary frequencies. Both TS′Cl,NO2 and TS′Cl,N(CH3)2
structures have lower imaginary frequencies than TSCl,NO2 and TSCl,N(CH3)2, whereas this is
not the case for the axially ligated acetonitrile structures.
Thus, we report here the effect of meso-substitution on the electronic and reactivity
properties of iron(IV)-oxo porphyrin cation radical models. We find little changes in
electronic configuration of the various CpdI reactants. However, fluorine substituents on the
meso-position of TPFPP entice favorable electrostatic interactions with the approaching
substrate and stabilize the epoxidation transition states. In the following we will try to
rationalize these observations.
149
5.4 Discussion
In this work we report a systematic set of epoxidations of substituted styrenes using four
different oxidants, namely 1Cl, 1NCCH3, 2Cl and 2NCCH3. During the substrate epoxidation
process two electrons are transferred from substrate to oxidant, which is reduced from
[FeIV
(O)(Por+•
)X] to [FeIII
(Por)X]. As a consequence, the electron affinity (EA) of the
oxidant and the ionization energy (IE) of the substrate should reflect the electron transfer
processes that take place. Indeed several experimental studies found a correlation between the
ionization energy of substrates and the rate constant of substrate epoxidation or sulfoxidation
(Lanucara, F.; Crestoni, M. 2011, Watanabe, Y.; Iyanagi, T. et al 1980, Watanabe, Y.;
Numata, T. et al 1981, Watanabe, Y.; Iyanagi, T. et al 1982). So far, no experimental studies
have been reported on the activation parameters of substrate sulfoxidation by synthetic
biomimetic iron(IV)-oxo porphyrin cation radical models, but the rate constants are expected
to correlate with the activation enthalpies. In earlier work we reported a systematic study of
olefin epoxidation by iron(IV)-oxo complexes and set up a model that predicts barrier heights
from known ionization potentials. Although that correlation still applies, we generalize it
further in this work using the data described here.
To understand the individual contributions of oxidant and substrate onto the reaction
process, i.e. transition states, we will investigate those separately. Let us start with a
discussion of the differences and comparison between 1Cl/2Cl and 1NCCH3/2NCCH3 in styrene
epoxidation reactions and their corresponding electron affinities. Thus, as shown in Figure
5.2 above the reactant complexes give little differences in geometric and electronic features
upon changing the equatorial ligand from Por to TPFPP.
We calculated EA values of 79.4 kcal mol–1
for 1Cl, 101.4 kcal mol–1
for 2Cl, 148.7 kcal
mol–1
for 1NCCH3 and 163.8 kcal mol–1
for 2NCCH3 in the gas-phase. Note here that despite the
fact that the valence orbital occupations and orbital shapes seem little influenced by the
150
nature of the meso-substituents of the porphyrin ring, actually the EA values increase by as
much as 15.1 – 22.0 kcal mol–1
. This observation is in excellent agreement with the
electrochemical studies of Fujii and co-workers who found similar trends (Takahashi, A.;
Kurahashi, T. et al 2011).
Since, the electrophilic reaction mechanism results in the electron transfer from substrate
into the oxidant orbitals, these differences in electron affinity of the four oxidants also affect
the subsequent reaction mechanisms and reactivities with substrates. The transition state
structures for these reactions, however, show dramatic differences in group spin densities as
displayed in Figure 5.7. Thus, generally the reaction proceeds with electron transfer from the
substrate into the half-filled a2u porphyrin orbital, whereby a decrease of spin density of the
porphyrin group is found. The electronic configuration of the TSCl,Z structures is very much
product-like with Por in the range from –0.29 to –0.47, whereas the TSAN,Z values are
reactant-like with considerably more spin density on the porphyrin ring.
Figure 5.7: UB3LYP/BS2//UB3LYP/BS1 calculated group spin density ranges for para-Z-
styrene epoxidation by 1Cl (top) and 1AN (bottom).
Z
ClZ
Fe = 1.43 – 1.64
O = 0.21 – 0.54
Sub = –0.37 – –0.53
Por = –0.47 – –0.29
Cl = –0.08 – –0.12
Fe = 1.07 – 1.40
O = 0.51 – 0.78
Sub = –0.02 – –0.85
Por = –0.16 – –0.79
NCCH3 = –0.03 – –0.04
151
Similarly, a polarization of the FeO spin density from almost equal oxygen and iron spin
densities in the reactants to dominant iron radical in TSCl,Z occurs, whereby Fe ranges from
1.43 – 1.64 and O ranges from 0.21 – 0.54. By contrast, Fe varies from 1.07 – 1.40 and O
ranges from 0.51 – 0.78 in the set of TSAN,Z structures. The axial ligand, therefore, has an
electronic effect on the charge and spin distributions in the rate determining transition states
and, in particular, an anionic axial ligand, such as chloride, polarizes the FeO biradical
towards the metal. Consequently, the anionic ligand will incur a “push”-effect on the metal-
oxo group and formally change it from an FeVO
–• configuration in the reactant state to
FeIV
O2–
in the transition state. By contrast, an iron(IV)-oxo oxidant with a neutral axial ligand
retains much larger radical character on the oxygen atom in the electrophilic transition states
and keeps the metal-oxo group in a formal FeVO
–• configuration. This is important as it
reduces the barrier heights of electrophilic reaction mechanisms.
Let us in the following look into the barrier heights of styrene epoxidation and elucidate the
enthalpic contributions to the relative values of the associated rate constants. Experimental
studies on substrate sulfoxidation by P450 enzymes indicated a correlation between rate
constant and ionization energy (IE) of the selected substrates (Watanabe, Y.; Iyanagi, T. et al
1980, Watanabe, Y.; Numata, T. et al 1981, Watanabe, Y.; Iyanagi, T. et al 1982). More
recent mass spectrometric and computational studies showed that this correlation also applies
to substrate epoxidation reactions (Lanucara, F.; Crestoni, M. 2011, Kumar, D.; Karamzadeh,
B. et al 2010). To test this relationship for the set of data studied here, we plot in Figure 5.8
the calculated barrier heights 2TSCl,Z and
2TSAN,Z as a function of the ionization energies of
the para-Z-styrene substrates. We calculated the trends using two sets of relative energies: In
the first we take the isolated reactants as a reference point, whereas in the second set a
reactant complex (RC) is used. DFT calculated ionization potentials at the same level of
152
theory as the transition states for the epoxidation reactions were used for consistency. Our
calculated IE values are close to those reported in the literature (Lide, D. 1996), however, as
not all IE values are experimentally known, we will use the calculated data in our analysis
here. As can be seen both series of para-Z-styrene epoxidation reactions give barriers that
correlate linearly with ionization potential, whether energies relative to isolated reactants or a
reactant complex is used. Interestingly, the two correlations that use energies relative to
isolated reactants have different slope and intercept, and indeed the two curves cross at an
ionization energy of about 9.40 eV. Therefore, [FeIV
(O)(Por+•
)NCCH3]+ in the gas-phase is a
better oxidant than [FeIV
(O)(Por+•
)Cl] in epoxidation reactions with substrates with ionization
energy below 9.40 eV, whereas it is reversed for substrates with higher ionization potential,
i.e. for para-CN-styrene and para-NO2-styrene. This observation is in excellent agreement
with experimental studies of Nam et al (Song, W.; Ryu, Y. et al 2005) who also found
improved reactivity of [FeIV
(O)(Por+•
)NCCH3]+ over [Fe
IV(O)(Por
+•)Cl] for the epoxidation
of para-substituted styrenes with large +
p Hammett value of the substituent. The study that
uses a reactant complex as a reference point, by contrast, does not find this change in
reactivity but instead produces almost parallel trends (Figure 5.8b). It may very well be,
therefore, that the reactant complexes are unrealistic models and do not compare well with
experiment. Thus, in a reactant complex the structure of oxidant and substrate is solvated by
solvent molecules and the interactions between all particles. In our reactant complex no
solvent molecules were included and as a consequence full freedom of optimization was
possible, which may have resulted in an experimentally unrealistic structure.
153
Figure 5.8: Styrene epoxidation barrier heights (E‡+ZPE) of
2TSX,Z plotted against the
ionization energy of the corresponding substrate. (a) Energies relative to isolated reactants.
(b) Energies relative to a reactant complex (RC).
To further understand the reactivity trends and the effect of substrate as well as oxidant on
the obtained barrier heights and, by extension, rate constants we set up a valence bond (VB)
curve crossing diagram for styrene epoxidation by iron(IV)-oxo porphyrin cation radical
systems, Figure 5.9. The VB curve crossing diagram displayed here is analogous to that
presented before for hydrogen atom abstraction reactions by iron(IV)-oxo oxidants (Shaik, S.
1981), but is further generalized to accommodate substrate epoxidation reactions.
This diagram starts on the bottom-left with the reactant configurations, i.e. styrene and
[FeIV
(O)(Por+•
)X]. The latter appears in two VB configurations, where we highlight the
valence -orbital electrons with dots. Thus, the electronic ground state of [FeIV
(O)(Por+•
)X] is
x2–y22 xz
2 yz
2 *xz
1 *yz
1 a2u
1 in both the quartet and doublet spin states and its wave
y = 1.33x - 4.94
R² = 0.88
y = 7.52x - 63.15
R² = 0.98
-15
-10
-5
0
5
10
4 5 6 7 8 9 10
Cl
ANE‡+ZPE
[kcal mol–1]
IE [eV]
(a)
y = 2.11x - 16.39
R² = 0.905
y = 2.42x - 13.97
R² = 0.96
-5
0
5
10
15
6 7 8 9 10
AN
ClE‡+ZPE
[kcal mol–1]
IE [eV]
(b)
154
function is labelled as r in Figure 5.9. The x2–y2 orbital is a nonbonding orbital in the plane
of the porphyrin ring. The perpendicular set of orbitals, xz/*xz and yz/*yz, are the bonding
and antibonding combinations of the metal 3dxz/3dyz with the oxygen 2px/2py atomic orbitals
and contain three electrons in each pair.
Figure 5.9: Valence bond curve crossing diagram for para-Z-styrene epoxidation by
iron(IV)-oxo porphyrin cation radical oxidants. Lewis structures give relevant valence-
orbitals with a dot.
In the C–O bond formation step in the epoxidation mechanism, one electron is transferred
from the substrate to the oxidant and it fills the a2u orbital with a second electron to give the
radical intermediate wave function I. In VB theory (Shaik, S. 1981), the reactant (r) and
product (I) wave functions cross each other and connect to excited states in the product and
reactants conformations, respectively. Thus, I* is an excited reactant state with an electronic
155
configuration that represents a C–O bond pair, a radical on the styrene group and a closed
shell a2u orbital. The two VB curves give an avoided crossing and a barrier height for C–O
bond formation, E‡. Based on a series of hydrogen abstraction reactions (Shaik, S. 1981,
Shaik, S.; Kumar, D. et al 2008, Latifi, R.; Bagherzadeh, L. et al 2009) it was shown that E‡
is proportional to the curve crossing energy (Ec) minus the resonance energy B via E‡ =
Ec – B. However, the curve crossing energy is proportional to the promotion gap (GH,r) or
excitation energy from the reactant to product configuration in the geometry of the reactants
so that the barrier height can be written as:
E‡ = f GH,r – B (1)
In a recent study we showed that the relationship in Equation 1 is also valid for substrate
epoxidation and sulfoxidation reactions and that the promotion gap GH,r reflects the excitation
energy or ionization potential of the substrate (Kumar, D.; Karamzadeh, B. et al 2010,
Kumar, D. Sastry, G. et al 2011).
A comparison of the VB structures for the r and I* configurations show that there is an
electron transfer from the iron(IV)-oxo group into the porphyrin a2u orbital reflecting an
electron excitation in CpdI (Eex,CpdI), which can be approximated with the electron affinity of
the oxidant. In addition, it can be seen that the electrons in the C=C bond are singlet paired in
the reactant ground state but triplet coupled in the excited state in the reactant geometry,
which implies a singlet-triplet energy gap in the C=C bond, EST. Thus, a singlet-triplet
energy gap in the C=C bond reflects the excitation energy of an electron in the -bond of the
C=C moiety, namely Eex,Z, which in its turn can be approximated with the ionization energy
of the substrate. Consequently, GH,r is proportional to EACpdI(X) + Eex,Z and so should the
barrier height of the epoxidation reaction. Of course, electron excitation from the -orbital of
156
the substrate, which is the HOMO, is also proportional to the ionization energy of the
substrate. Indeed, the plot in Figure 5.8 confirms a relationship between barrier height and
ionization potential as predicted by the VB diagram in Figure 5.9.
For the reverse reaction, that is from radical intermediates to reactants, the epoxidation
barrier height (E‡
rev) is equal to E‡ plus the exothermicity to form radical intermediates
(Eri). In VB the reverse barrier is proportional to the promotion gap in the radical
intermediates, i.e. I to p* or GH,p. The VB structures on the radical intermediate side of
Figure 5.9 show that the excitation energy GH,p corresponds to the formation energy of the C–
O bond plus the electron transfer from the substrate into the a2u orbital. The VB structures in
Figure 5.9 indicate that the electrons in the C–O bond are singlet coupled in I but triplet
coupled in p*, which implies that a singlet-triplet excitation in the C–O bond has occurred.
The singlet-triplet excitation in the C–O bond refers to the bond breaking of the C–O bond. In
earlier work we showed that the C–O bond formation energy is proportional to a H–O bond
formation energy, so that it can be mimicked with BDEOH as defined as the reaction enthalpy
for Equation 3 (Kumar, D.; Karamzadeh, B. et al 2010).
To test whether the VB curve crossing diagram can predict barrier heights from empirical
values, we calculated the gas-phase electron affinities of all four oxidants from Figure 5.1 and
the -* excitation energies and ionization energies of all para-Z-styrene substrates and the
results are given in Table 4.1. We calculated the electron affinity of the oxidant, the -*
excitation energy in the para-Z-styrene substrates and the promotion gap via Equation 2.
Subsequently, we used these values to estimate E‡
VB for all substrates. Although these VB
calculated barrier heights are on average within –0.05 kcal mol–1
of the DFT calculated ones,
actually the standard deviation is quite large (about 2 kcal mol–1
). This implies that the model
has short-comings, which we address with a slightly modified model as described below.
157
GH,r = 2 (EACpdI(X) + Eex,Z) (2)
[FeIV
(O)(Por+•
)X] + H• → [Fe
IV(OH)(Por)X] (3)
Thus, the forward reaction barrier is dependent on the intrinsic chemical properties of the
substrate, i.e. the ionization energy, whereas the reverse reaction barrier depends on the
variables of the oxidant that is the BDEOH value. We calculated gas-phase values of BDEOH =
87.5 kcal mol–1
for [FeIV
(O)(Por+•
)Cl] and BDEOH = 82.8 for [FeIV
(O)(Por+•
)NCCH3]+,
respectively (de Visser, S.; Tahsini, L. et al 2009). This energy difference narrows to almost
equal values in a dielectric constant of = 5.7: BDEOH = 82.3 and 81.3 kcal mol–1
for
[FeIV
(O)(Por+•
)X], X = Cl– or NCCH3, respectively.
The VB diagram of Figure 5.9 shows that the transition state described from reactants to
intermediates, i.e. forward, is proportional to EACpdI(X) + Eex,Z, whereas the transition state in
the reverse reaction, i.e. backwards, is proportional to the BDEOH value of the oxidant. Thus,
the location of the barrier on the potential energy surface will determine whether the barrier
correlates with EACpdI(X) +Eex,Z or BDEOH. So, for a substrate epoxidation reaction where the
transition state has a very reactant-like electronic configuration, i.e. an early transition state,
very little electron transfer from substrate to oxidant has taken place and consequently the
reactant wave function dominates in the transition state. As a result, an early transition state
should be proportional to BDEOH. On the other hand, a late transition state has an electronic
configuration closely resembling the radical intermediate state due to a significant amount of
electron transfer that has taken place already. Therefore, late transition states, based on the
VB diagram in Figure 5.8 above should correlate with the electron affinity of the oxidant.
Indeed the group spin density of the epoxidation transition state with para-NO2-styrene as a
substrate shows a only a small amount of electron transfer from substrate to oxidant and
consequently the charge transfer from substrate to oxidant (QCT) is almost zero. This implies
158
that the epoxidation barriers for para-NO2-styrene will be early on the potential energy
surface and little electron transfer has taken place. As reasoned above, these barrier heights
should, therefore, correlate with BDEOH and not with EACpdI(X) + Eex,Z. To test this we display
in Figure 5.10 the correlation of para-NO2-styrene epoxidation with BDEOH for 1X/2X (X =
Cl– and NCCH3). As can be seen for this set of four data points a linear correlation is found
between the epoxidation barrier of para-NO2-styrene and the corresponding BDEOH value of
the oxidant.
Figure 5.10: Correlations of (a) Epoxidation barrier height of para-NO2-styrene with
BDEOH. (b) Epoxidation barrier height of para-N(CH3)2-styrene with EACpdI(X).
The group spin densities of 2TSX,N(CH3)2, by contrast to those found for
2TSX,NO2 show
considerably larger amount of spin density on the substrate part of the structure. At the same
time the spin density on the porphyrin ring has dropped due to electron transfer from
substrate to oxidant. Thus, the 2TSX,N(CH3)2 structures are electronically late and resemble
0
2
4
6
8
10
80 85 90 95 100
E‡+ZPE
[kcal mol–1]
BDEOH [kcal mol–1]
Z = NO2(a)
-25
-20
-15
-10
-5
0
5
10
75 95 115 135 155 175
E‡+ZPE
[kcal mol–1]
EA [kcal mol–1]
Z = N(CH3)2
(b)
159
product type conformations. Because of that, in 2TSX,N(CH3)2 the electron transfer has taken
place already (or at least most of it) and, consequently, the barrier height will be driven by the
differences in electron affinity of the individual oxidants. To test this, we plot in Figure
5.10(b) the correlation between epoxidation barrier height of para-N(CH3)2-styrene
epoxidation by 1X/2X oxidants as a function of the electron affinity of the oxidant. We find a
linear correlation between electron affinity and barrier height as predicted by the VB diagram
in Figure 5.8.
As follows from the correlations depicted in Figure 5.10, the amount of electron transfer
from substrate to oxidant determines whether a transition state correlates with either the sum
of EACpdI(X) + Eex,Z or with BDEOH instead. Thus, we extracted the degree of charge transfer
from substrate to oxidant (QCT) from the UB3LYP/B2//UB3LYP/B1 calculations and
summarize these values for the 2TSX,Z transition states in Table 5.1. We now define the
variable that describes the position of the transition state on the potential energy profile
through Equation 4 and link to the degree of charge transfer (QCT), BDEOH and EACpdI(X) +
Eex,Z.
= QCT BDEOH + (1 – QCT) (EACpdI(X) + Eex,Z) (4)
Essentially, is proportional to BDEOH when the charge-transfer is maximal, i.e. an early
transition state, but it is proportional to EACpdI(X) + Eex,Z for a late transition state with a small
value of QCT. We calculated for all transition states 2TSX,Z, X = Cl
–/NCCH3 and para-Z-
styrene as substrates, and a plot of all data is given in Figure 5.11. Thus, the full set of
transition states, irrespective of the axial ligand of the oxidant, fits a linear correlation.
Consequently, we described the trend in epoxidation barriers for early as well as late
160
transition states as a function of intrinsic chemical properties of the oxidant and substrate and
the amount of charge transfer in the TS during the reaction process.
Figure 5.11: Correlation between epoxidation barrier height of all data for 1X (X = Cl–,
NCCH3) with parameter
Table 5.1. Substrate chemical properties and charge-transfer (QCT) values in the
transition states.
Z IEZ a Eex,Z
a QCT,Cl
b QCT,AN
c
H 9.05 3.82 0.10 0.25
F 9.05 3.81 0.07 0.24
Cl 8.87 3.70 0.08 0.16
CH3 8.59 3.75 0.12 0.32
t-Bu 8.43 3.76 0.12 0.42
CN 9.46 3.46 0.01 0.10
NO2 9.70 2.88 –0.03 –0.01
OCH3 8.03 3.73 0.16 0.42
NH2 7.55 3.61 0.22 0.55
N(CH3)2 7.05 3.57 0.27 0.63
a In eV.
b Charge transfer in
2TSCl,Z.
c Charge transfer in
2TSAN,Z.
y = 2.47x + 136.46
R² = 0.86
0
50
100
150
200
-15 -10 -5 0 5 10 15E‡+ZPE [kcal mol–1]
[kcal mol–1]
161
In summary, we show here that the barrier height of an epoxidation reaction by a metal-oxo
oxidant is dependent on the electron donating/withdrawing character of the substituents of the
substrate. Thus, substrates with high electron donating power such as para-N(CH3)2-styrene
result in low barrier and rapid electron transfer from substrate to oxidant well before the
epoxidation barrier height has overcome. On the other hand, electron withdrawing groups
such as para-NO2-styrene result in much later electron transfer from substrate to oxidant and
now the height of the barrier is determined by the strength of the C–O bond that is formed.
5.5 Conclusion
In this work we report a systematic computational study into substrate epoxidation by four
iron(IV)-oxo porphyrin cation radical models. We investigated the effect of the axial ligand,
chloride versus acetonitrile, as well as the equatorial ligand, Por versus TPFPP. It is shown
that the substituents on the porphyrin ring can guide substrate binding through electrostatic
interactions with halide atoms, which lowers the barrier heights. A neutral axial ligand leads
to displacement of the metal from the plane through the porphyrin ring and results in different
orbital interactions between metal and porphyrin ring as compared to systems with an anionic
ligand. This has profound effects on the electron affinity of the oxidant and the subsequent
reactivity patterns. Finally, we investigated para-Z-styrene epoxidation by four iron(IV)-oxo
porphyrin cation radical models and highlight the differences in reactivity for olefins with
electron withdrawing versus electron donating substituents. In particular, it is shown that
electron donating substituents give early transition states and lower reaction barriers than
substrates with electron withdrawing substituents.
162
CHAPTER 6
PROJECT FOUR
163
Drug metabolism by cytochrome P450 enzymes:
What distinguishes the pathways leading to
substrate hydroxylation over desaturation?4
ABSTRACT
Cytochrome P450 enzymes are highly versatile biological catalysts in our body that react
with a broad range of substrates. Key functions in the liver include the metabolism of drugs
and xenobiotics. However, these processes often lead to reactive (toxic) metabolites that are
harmful. Understanding the biodegradation of molecules by P450 enzymes is therefore
important for pharmaceutical purposes and may assist in developing better medication. One
particular metabolic pathway that is poorly understood relates to the P450 activation of
aliphatic groups leading to either hydroxylation or desaturation pathways. We have done the
first combined density functional theory and quantum mechanics/molecular mechanics study
on the factors that determine the regioselectivity of aliphatic hydroxylation over desaturation
of compounds by P450 isozymes. The calculations establish multistate reactivity patterns on
competing doublet and quartet spin states, whereby the product distributions differ on each of
the spin state surfaces; hence we find spin-selective product formation. We analyzed the
electronic and thermochemical factors that determine the bifurcation pathways and have
established a model that predicts the regioselectivity of aliphatic hydroxylation over
4 Li Ji, Abayomi S. Faponle, Matthew G. Quesne, Mala A. Sainna, Jing Zhang, Alicja Franke, Rudi van Eldik, Weiping Liu and Sam P. de
Visser. “Drug metabolism by cytochrome P450 enzymes: What distinguishes the pathways leading to substrate hydroxylation over
desaturation?”Chem. Eur. J. 2015, 21, 1-11.
164
desaturation pathways from valence bond and molecular orbital theories. Thus, the difference
in energy of the C–H bond orbital versus that of the substrate radical orbital in the radical
intermediate determines the degree of desaturation products. In addition, we have identified
environmental effects of the substrate binding pocket that affect the regioselectivities. These
studies imply that bioengineering P450 isozymes for desaturation reactions will have to
include modifications in the substrate binding pocket to restrict the hydroxylation rebound
reaction.
6.1 Introduction
The cytochrome P450s (P450s) are key drug metabolism enzymes in the human body
(Sono, M.; Roach, M. P.; et al 1996, Groves, J. T. 2003, Ortiz de Montellano, P. R., 2004,
Denisov, I. G.; Makris, T. M.; et al 2005, Kadish, K. M.; Smith, K. M.; et al 2010, Ortiz de
Montellano, P. R. 2010), and are widely distributed in living organisms, such as eukaryotes,
but also in plants, fungi and bacteria. As of January 2014, more than 20,000 different
sequences had been identified, making the P450s one of the major biological enzyme classes
(Nelson, D. R. 2009). In the human body, P450s are predominantly found in the liver, where
they catalyze the biodegradation of xenobiotics but are also involved in the biosynthesis of
hormones (Guengerich, F. P. 2001, Posner, G. H.; O’Neill, P. M. 2004, Munro, A. W.;
Girvan, H. M.; et al 2007). As such the P450s have multiple functions in the body, which also
shows in their versatility on substrate activation. For instance, the P450BM3 subclass
hydroxylates long-chain fatty acids (Gelb, M. H.; Heimbrook, D. C.; et al 1982, Atkins, W.
M.; Sligar, S. G. 1987), and has a substrate binding pocket that accommodates long and thin
substrates. By contrast, P450cam is a bacterial isozyme that regioselectively hydroxylates
camphor at the carbon-5 position and has a fit-for-purpose substrate binding pocket
(Ruettinger, R. T.; Wen, L. P.; et al 1989, Davydov, D. R.; Bon Hoa, G. H.; et al 1999). Key
drug metabolizing P450 isozymes include P4502D6, which has a more open substrate
165
binding pocket and can activate a large range of different substrate shapes and sizes. As P450
isozymes are highly efficient in substrate activation, they have gained interest from the
biotechnology industry for the regioselective and/or stereospecific activation of substrates
(O'Reilly, E.; Koehler, V.; et al 2011, Grogan, G. 2011). Despite decades of extensive
scientific research into P450 chemistry, there are still major gaps in our understanding of its
reactivity and activity, in particular related to the metabolism of drug molecules and
xenobiotics, which is important for drug development and the pharmaceutical industry.
The name cytochrome P450 derives from the fact that the CO bound heme complex has an
absorption band at 450 nm (Omura, T.; Sato, R. 1962). Their heme active site is linked to the
protein via a thiolate linkage of a cysteinate residue covalently bound to the iron center
(Poulos, T. L.; Finzel, B. C.; et al 1985, Schlichting, I.; Berendzen, J.; et al 2000, Auclair, K.;
Moënne-Loccoz, P.; et al 2001), while the substrate binding pocket varies in shape and size
for the different P450 isozymes. As an example we show in Figure 5.1 the active site of
P4502C19 with substrate warfarin bound as taken from the 4GQS protein databank (pdb) file
(Reynald, R. L.; Sansen, S.; et al 2012). The heme is linked to the protein through an iron-
thiolate linkage with Cys435 and trans to this ligand the substrate (warfarin) binds.
166
Figure 6.1: Extract of the active site of P4502C19 as taken from the 4GQS pdb file.
The active species of P450 enzymes is an iron(IV)-oxo heme cation radical species called
Compound I (Cpd I) (Rittle, J.; Green, M. T. 2010). Due to its transient nature, Cpd I is short-
lived and difficult to study with experimental means and as a consequence many details of its
catalytic mechanism with substrates are unknown. The most common reaction mechanism
performed by the P450s is aliphatic hydroxylation of substrates (Groves, J. T. 2005,
Watanabe, Y.; Nakajima, H.; et al 2007), which was experimentally revealed as a stepwise
process leading to alcohol product complexes (Groves, J. T.; McClusky, G. A. 1976).
Computational modelling confirmed the general mechanism and showed that the oxidation
heme
Cys435
warfarin
167
actually contains two close-lying spin states that lead to two-state reactivity patterns, each
with an individual rate constant of the reaction (Kamachi, T.; Yoshizawa, K. 2003, Shaik, S.;
Kumar, D.; et al 2005). Moreover, the two-state reactivity pattern also predicted that the
lifetime of the radical intermediate could lead to a bifurcation of the potential energy profile,
thereby generating, for instance, rearrangement patterns on one spin-state surface but not on
the other (De Visser, S. P.; Ogliaro, F.; et al 2001).
A key reaction mechanism in P450 chemistry with big relevance to drug metabolism is the
desaturation of aliphatic groups to form olefins. As shown in Scheme 6.1 desaturation is a
reaction mechanism that shows similarities to aliphatic hydroxylation, whereby both
processes start with an initial hydrogen atom abstraction. However, thereafter the reaction
bifurcates into two possible product channels: one giving OH rebound to form alcohol
products (aliphatic hydroxylation) and the other leading to a second hydrogen atom
abstraction to give an olefin and water (desaturation). The origin and consequences of the
bifurcation pathways in P450 catalysis are currently unknown and no predictive model exists
that gives product distributions as a function of substrate chemical properties.
Several drug molecules have been shown to produce desaturation metabolites after a
reaction with P450 enzymes. Namely, valproic acid, an antiepileptic drug, is converted by
P450 enzymes to form both alcohol and olefin products (Rettie, A. E.; Rettenmeier, A. W.; et
al 1987, Rettie, A. E.; Boberg, M.; et al 1988, Sadeque, A. J. M.; Fisher, M. B.; et al 1997,
Wen, X.; Wang, J.-S.; et al 2001, Gunes, A.; Bilir, E.; et al 2007). Another example is
ethylcarbamate (Loch, J. M.; Potter, J.; et al 1995, Forkert, P.-G.; Lee, R. P. 1997, Lee, R. P.;
Parkinson, A.; et al 1998, Guengerich, F. P. 2008), that in a reaction with P4502E1 enzymes
leads to a small amount of vinyl carbamate as a precursor to its epoxide (Guengerich, F. P.;
Kim, D. H. 1991). The latter is considered the ultimate carcinogen (Leithauser, M. T.; Liem,
A.; 1990) and its biosynthesis requires an essential desaturation step. Over the years a range
168
of substrates of P450 isozymes have been identified that are desaturated to form olefins,
including lauric acid (Guan, X.; Fisher, M. B.; et al 1998), but also a step in the biosynthesis
of ergosterol includes a desaturation (Kelly, S. L.; Lamb, D. C.; 1997).
Clearly, the mechanistic pathways surrounding drug metabolism by the P450s produce a
large versatility of product patterns and currently it is not clear why certain substrates lead to
desaturation, whereas others solely give hydroxylation products. To gain insight into the
origin of the regioselectivity of desaturation versus hydroxylation processes by P450
enzymes, we decided to do a comprehensive computational study that includes density
functional theory (DFT) modelling, quantum mechanics/molecular mechanics (QM/MM) and
valence bond (VB) theory. The overall study gives an electronic structure description on the
metabolism of a variety of drug molecules and enables us to provide a predictive pattern that
anticipates desaturation versus hydroxylation product distributions.
Scheme 6.1: (a) Competitive hydroxylation and desaturation metabolism pathways of drug
molecules by P450 Cpd I. (b) Reaction products observed for valproic acid and
ethylcarbamate.
169
Our studies explain the experimentally observed bifurcation reactions and may be used to
develop computer-aided toxic risk assessment of potential carcinogens resulting from
desaturation reactions catalyzed by P450 isozymes.
6.2 Methods
Our work uses two main computational approaches, namely DFT on model complexes and
QM/MM on a full P450 isozyme.
6.2.1. DFT model calculations
As in previous studies of our groups (de Visser, S. P.; Tan, L.-S. 2008, Ji, L.; Zhang, J.; et
al 2014), Cpd I of P450 was modeled as a six-coordinated iron(IV)-oxo porphyrin (Por)
cation radical complex, [FeIV
(O)(Por+
)(SH)]. For simplicity, the heme is replaced by
protoporphyrin IX, whereby all side chains are abbreviated with hydrogen atoms and the
axial cysteinate ligand is changed to thiolate. All geometries were optimized using the
unrestricted hybrid B3LYP density functional (Becke, A. D. 1993, Lee, C.; Yang, W.; et al
1988) in combination with an LACVP basis set on iron and 6-31G on the rest of the atoms
(Hay, P. J.; Wadt, W. R. 1985, Hehre, W. J.; Ditchfield, K.; et al 1972); basis set BSI. A
subsequent analytical frequency calculation was run to confirm that all ground states had no
imaginary frequencies, and all transition states had a single one for the correct mode. The
computed vibrational frequencies were then used further for quantifying the zero-point
energy (ZPE), and enabled us to calculate thermal and entropic corrections to the free energy
at 298.15 K and 101.325 kPa. More accurate energies were determined by single-point
calculations with the SDD basis set on iron coupled to 6-311++G** on the rest of the atoms;
basis set BSII. To test the effect of the basis set on the optimized geometries, we calculated a
full potential energy profile of 1-butene epoxidation by Cpd I using different basis sets.
However, a single point calculation with a triple- quality basis set on all atoms performed on
a geometry that was optimized with either a double- or a triple- quality basis set did only
170
give minor differences in the relative energies to within 0.1 kcal mol–1
of each other (Kumar,
D.; Karamzadeh, B.; 2010), hence the former procedure was used in this work.
Bulk protein environment was simulated with the polarized continuum-solvation model
with a dielectric constant of = 5.62. Dispersion interactions were added to the energy at the
UB3LYP-D3 level of theory with zero damping (Grimme, S. 2006). All calculations were
carried out using the Gaussian 09 program package (Frisch, M. J. 2010).
As density functional theory occasionally struggles with the correct description of spin
state ordering of transition metal complexes (de Visser, S. P.; Quesne, M. G.; et al 2014), we
tested the effect of the density functional method on the spin state ordering and relative
energies of the rebound versus desaturation reactions. We did full geometry optimizations at
the UBLYP/BSI (Lee, C.; Yang, W.; et al 1988, Becke, A. D. 1988), UB3PW91/BSI (Becke,
A. D. 1993, Perdew, J. P.; 1992), and TPSS/BSI (Tao, J.; Perdew, J. P.; et al 2003) levels of
theory. However, no change in spin state ordering was observed and only minor differences
in the relative energies of the barrier heights were obtained. Therefore, the same qualitative
picture is obtained with alternative DFT methods and we will focus on the B3LYP results
only.
6.2.2. QM/MM calculations
To support the DFT calculations and gain insight into whether the model would be
applicable to realistic systems, we did a further set of calculations using an actual P450
isozyme that is known to desaturate its natural substrate. OleTJE was recently identified as a
P450 isozyme that converts long chain fatty acids into terminal olefins (Rude, M. A.; Baron,
T. S.; et al 2011). We investigated the pathways of -hydroxylation and desaturation of
eicosanoic acid by the P450 isozyme OleTJE using QM/MM methods. The starting
coordinates for the calculations were taken from crystal structure data (Belcher, J.; McLean,
K. J.; et al 2014). We followed the set-up procedures of the QM/MM system as reported and
171
benchmarked before, which we will briefly summarize here (Kumar, D.; Thiel, W.; et al
2011, Quesne, M. G.; Latifi, R.; et al 2014).
Starting from the crystal structure coordinates, hydrogen atoms and oxo group were added
to the structure using the PDB2PQR program package (Dolinsky, T. J.; Czodrowski, P.; et al
2007) and the active site was manually modified from a resting state iron(III)-water(heme)
complex into an iron(IV)-oxo heme cation radical (Cpd I) structure. We did a careful analysis
of all acidic and basic amino acid side chains for their protonation states, whereby we chose
all glutamic and aspartic acid residues to be deprotonated at the carboxylic acid group and all
arginine and lysine residues protonated. All histidine groups were visually inspected for
proton donor and acceptor possibilities. The histidine residues His120 and His210 were doubly
protonated, whereas the other six (His85, His92, His222, His259, His325 and His363) were singly
protonated. The overall structure was neutralized with counter-ions (Mg2+
and Cl–). Solvent
water (with sphere of radius of 35 Å) was added to this structure, and equilibrated, followed
by a molecular dynamics minimization and heating procedure to 298 K of the full structure
using the CHARMM force field (Brooks, B. R.; Bruccoleri, R. E.; et al 1983). The thus
obtained QM/MM model had a total number of 32,992 atoms including 8,739 TIP3P water
molecules.
Three snapshots from the MD simulation were selected as starting points for the QM/MM
calculations at time intervals of 300, 400 and 500 ps (Sn300, Sn400 and Sn500). In these
QM/MM calculations the inner core including all residues and water molecules within a
radius of 8 Å from the heme iron were fully geometry optimized, whereas all other atoms
were fixed in the MD simulated coordinates. The iron(III)-hydroxo species was geometry
optimized in the doublet and quartet spin states and a frequency calculation on the QM region
established these structures as stationary points.
172
QM/MM calculations employed the Turbomole program package (Ahlrichs, R.; Bär, M.;
et al 1986) for the calculations of the QM region and Charmm as implemented in DL-Poly
(Smith, W.; Forester, T. R. 1996) for the MM region all linked through the ChemShell
interface (Sherwood, P. et al 2003). For consistency with the DFT studies reported above, we
selected the UB3LYP (Becke, A. D. 1993, Lee, C.; Yang, W.; et al 1988) density functional
for the QM region in combination with the Turbomole built-in basis set SVP, BSIII
(Ahlrichs, R.; Bär, M.; et al 1989). The QM region contained the iron(IV)-oxo group, the
heme (without side chains), thiolate for the axial cysteinate group and the butanoic acid group
of the substrate. To improve the energetics, single point calculations were performed with a
Wachters all electron basis set on iron and 6-311+G** on the rest of the atoms; basis set
BSIV (Wachters, A. J. H. 1970). Electronic embedding procedures were implemented
(Bakowies, D.; Thiel, W. 1996), thereby taking into consideration to electronic interactions of
electrons in the QM region with charges in the MM region.
Scheme 6.2 displays the QM region selected in the QM/MM calculations. As the substrate
has a terminal carboxylate group that is linked through a salt bridge with an arginine residue,
we kept the methylguanidinium component of Arg245 in the QM region and terminated the
substrate at the sixth carbon atom. All side chains of the heme were in the MM region and
cysteinate was abbreviated by methylmercaptane. Subsequently, we investigated hydrogen
atom abstraction from the C position of the substrate and the pathways leading to
desaturation and hydroxylation.
173
Scheme 6.2: Atoms in the QM region of the QM/MM calculation. Wiggly lines represent the
cuts between the QM and MM regions.
6.3 Results and Discussion
6.3.1. DFT model reactions
Before going into detail of the mechanism of drug metabolism by P450 enzymes, let us
first go through a typical reaction mechanism of substrate activation leading to desaturation
and hydroxylation products by taking ethylcarbamate (EC) as an example. Scheme 6.3
displays the calculated reaction mechanisms of EC activation by Cpd I of P450 and the
bifurcation mechanism leading to desaturation and hydroxylation products.
Scheme 6.3: Reaction Mechanism of Ethyl Carbamate Activation by Cpd I of P450.
174
Figure 6.2: Potential energy profile of ethylcarbamate activation by 4,2
Cpd I of P450 as
calculated with DFT. Energies are given in kcal mol–1
and are calculated at
UB3LYP/BSII//UB3LYP/BSI level of theory with ZPE and solvent corrections included.
Values in parenthesis are free energies in solvent. Optimized geometries give bond lengths in
angstroms and the imaginary frequency in the transition states in wave numbers.
We calculated the reaction between Cpd I and ethylcarbamate, as shown in Figure 6.2,
starting from a reactant complex (4,2
REC), which exists in close-lying doublet and quartet spin
states, both with electronic configuration x2–y22 *xz
1 *yz
1 a2u
1. Attack of the oxo group of
Cpd I on the substrate leads to a hydrogen atom abstraction via a transition state 4,2
TSH,EC to
form a radical intermediate (4,2
IH,EC) with electronic configuration x2–y22 *xz
1 *yz
1 a2u
2
Rad1. During the lifetime of the radical intermediate it can isomerize to structures
4,2ID,EC,
whereby the C–H group of the substrate forms a hydrogen bonding interaction with the
oxygen atom of the OH group as a precursor to the second hydrogen atom abstraction step via
4TSH,EC [2TSH,EC]
i1179 [i1208] cm–1
rCH = 1.471 [1.370]rOH = 1.112 [1.157]
4ID,EC [2ID,EC]
4IR,EC [2IR,EC]4IH,EC [2IH,EC]
rHO = 2.420 [2.410]rOH = 0.982 [0.982]
rCH = 2.588 [2.479]rOH = 0.984 [0.985]
rCO = 3.565 [3.568]rOH = 0.982 [0.982]
175
transition state 4,2
TSD,EC to form desaturated products 4,2
PD,EC. The alternative mechanism
from 4,2
IH,EC gives an isomerization to form the pro-hydroxylation radical intermediate
4,2IR,EC, where the radical is aligned to the oxygen atom for rebound via transition state
4,2TSreb,EC to give the alcohol product complexes
4,2PA,EC. Our calculations were focused on
establishing whether isomerization barriers exist from IH to ID and IR, and what the factors
are that determine the height of these barriers and consequently the regioselectivity of the
reaction.
As shown in Figure 6.2, the hydrogen atom abstraction step is endergonic by G = 19.4
(17.9) kcal mol–1
in the quartet (doublet) spin states, respectively. Addition of dispersion
correction to the calculated relative energy lowers 4,2
TSH,EC in energy by about 4 kcal mol–1
,
which is in accord with analogous previous findings for transition metal complexes
(Lonsdale, R.; Harvey, J. N.; et al 2010). The hydrogen atom abstraction transition states are
characterized by an almost linear O--H--C configuration and a large imaginary frequency
(HS: i1179 cm–1
; LS: i1208 cm–1
), and consequently a considerable kinetic isotope effect
(KIE) for the reaction may be expected (De Visser, S. P. 2006).
We then investigated the OH rebound to form alcohol products and the second hydrogen
atom abstraction to give desaturation products. We propose this to happen (Scheme 6.2) via
an initial isomerization of the radical intermediate from 4,2
IH,EC to 4,2
IR,EC, which sets up the
molecular orientation for the OH rebound leading to 2-hydroxyethyl-carbamate product.
Similarly to previous studies (Shaik, S.; Kumar, D.; et al 2008), a small but significant
rebound barrier (G = 4.9 kcal mol–1
) is found on the quartet spin surface, while the doublet
spin pathway is barrier free. Dispersion corrections lower the rebound barriers by about 1
kcal mol–1
, which is a significant lesser change than that found for the hydrogen atom
abstraction step. The radical rebound pathway from 4,2
IH,EC is exergonic by G = 58.8 (55.8)
kcal mol–1
on the quartet (doublet) spin state surfaces as expected for the formation of a
176
stable alcohol product complex. The alternative isomerization pathway from 4,2
IH,EC, leads to
a substrate rotation to form 4,2
ID,EC, whereby the oxygen atom of the OH group is aligned
with a hydrogen atom of substrate to enable a second hydrogen atom abstraction. Note that
the difference in orientation between 4,2
IR,EC and 4,2
ID,EC are similar to those reported for a
synthetic manganese-hydroxo porphyrin complex (Hull, J. F.; Balcells, D.; et al 2010). The
doublet spin radical intermediate 2ID,EC collapses to the desaturation product complex
virtually barrier-free, whereas the quartet spin complex 4ID,EC encounters a large activation
barrier (4TSreb,EC) of G = 17.8 kcal mol
–1 instead. Similarly to the above mentioned rebound
process, the activation barrier only decreases by 1.5 kcal mol–1
after dispersion corrections
are added. The large 4TSD,EC barrier implicates that desaturation is an unlikely process on the
quartet spin state surface. Moreover, the difference in kinetics between the quartet and
doublet spin desaturation routes suggests that the doublet spin state surface is the only viable
pathway responsible for vinyl carbamate formation. Consequently, the regioselectivity of
desaturation versus hydroxylation is a spin-selective reaction process, whereby the high-spin
pathway gives solely hydroxylation products whereas a mixture of products may be expected
on the low-spin surface instead.Experimental reports on ethylcarbamate activation by P450
isozymes reveal only a small proportion of biotransformation leading to desaturation
reactions (Guengerich, F. P.; Kim, D. H. 1991). The DFT calculations presented in Figure
6.2, therefore, are in line with these experimental product distributions of hydroxylation
versus desaturation, with a small thermodynamic and kinetic preference in favor of substrate
hydroxylation. Moreover, when we look at the driving force for desaturation versus
hydroxylation, both pathways on the low-spin surface have similar exergonicity, i.e. G =
−48.0 vs −47.7 kcal mol–1
. As such, P450s should exhibit mixed hydroxylase/desaturase
activity toward EC as indeed observed experimentally. To find a general trend in desaturation
versus hydroxylation regioselectivity preferences, we did a further set of calculations using
177
alternative substrates that should give the chemical extremes of the two processes. In Figure
6.3 we display studies on desaturation versus hydroxylation mechanisms using valproic acid
(VA), ethane (ET) and dihydroanthracene (DHA) as substrates. Thus, experimentally, DHA
in a reaction with P450 enzymes is known to lead exclusively to anthracene products,
whereas ethane is leads solely to alcohol products instead. For all chemical systems the full
potential energy profile like the one shown in Figure 6.2 is calculated on both spin state
surfaces and mechanisms are analogous to those reported above. Thus, for valproic acid as a
substrate the DFT calculations predict mixed desaturase/hydroxylase activity on a low-spin
surface. Key optimized geometries of the hydrogen atom abstraction transition states are
given in Figure 6.3. In general, all substrates reveal the same catalytic mechanism as
calculated above and previously reported in the literature (Kamachi, T.; Yoshizawa, K. 2003,
Shaik, S.; Kumar, D.; et al 2005, Ogliaro, F.; Harris, N.; et al 2000, de Visser, S. P.; Kumar,
D.; et al 2004, Ji, L.; Schüürmann, G. 2013) with a rate determining hydrogen atom
abstraction leading to a radical intermediate. Differences, however, are found for the relative
energies for radical rebound versus desaturation.
Figure 6.3: Optimized geometries of hydrogen atom abstraction transition states of VA, ET
and DHA by 4,2
Cpd I of P450 as calculated with DFT. Bond lengths are in angstroms, the
imaginary frequency is in wave numbers and (free) energies are given in kcal mol–1
and are
4TSH,VA [2TSH,VA]
i1560 [i1273] cm–1
rCH = 1.358 [1.248]rOH = 1.184 [1.284]
4TSH,ET [2TSH,ET] 4TSH,DHA [2TSH,DHA]
i1683 [i1099] cm–1
rCH = 1.402 [1.322]rOH = 1.150 [1.193]
i1505 [i1655] cm–1
rCH = 1.282 [1.246]rOH = 1.300 [1.346]
E+ZPE = 13.1 (12.1)G = 15.0 (14.4)
E+ZPE = 15.8 (13.7)G = 19.0 (17.1)
E+ZPE = 7.8 (6.9)G = 10.0 (9.3)
178
calculated at UB3LYP/BSII//UB3LYP/BSI level of theory with ZPE and solvent corrections
included.
Table 6.1. Free energies of activation of hydrogen atom abstraction, rebound and
desaturation barriers.a
Substrate TSH b TSreb
c TSD
d
LS:
Ethylcarbamate 17.9 0.0 0.0
Valproic acid 14.4 0.0 0.0
Ethane 17.1 0.0 14.7
DHA 9.3 0.0 0.0
HS:
Ethylcarbamate 19.4 4.9 17.8
Valproic acid 15.0 1.3 23.0
Ethane 19.0 1.7 35.6
DHA 10.0 0.8 4.2
a G values at 298 K in kcal mol
–1.
b G relative to R.
c G relative to IH.
d G relative to
ID.
On the quartet spin state surface a small but significant radical rebound barrier exists (0.8 –
4.9 kcal mol–1
), whereas all low-spin rebound pathways are essentially barrier free. Details of
the rebound and desaturation barriers for all substrates are given in Table 6.1.
By contrast to the radical rebound process, the desaturation barriers (TSD) show
considerably more variation in relative energies. Similarly to the ethylcarbamate mechanism
from Figure 6.2, a substantial desaturation barrier of 4.2 – 35.6 kcal mol–1
is required for
valproic acid, ethane and dihydroanthracene desaturation on the quartet spin state. Note as
well that there is a significant desaturation barrier on the low-spin pathway for ethane, which
is further evidence that desaturation is an unlikely process for a substrate like ethane.
179
Although, the barrier heights displayed in Table 6.1 implicate competitive hydroxylation
and desaturation on the doublet spin state surface, there appears to be a clear preference for
radical rebound on the high-spin surface. This is similar to previous studies on the
regioselectivity of substrate epoxidation versus suicidal complex formation in olefin
activation by P450 isozymes that was predicted to only lead to by-products on the high-spin
surface (De Visser, S. P.; Ogliaro, F.; et al 2001). It should be noted, however, that the
driving force for hydroxylation is larger for EC, VA and ethane as substrates, whereas it is
favorable for desaturation for DHA. Thermodynamically, therefore, DHA should
preferentially undergo the desaturation pathway, in accord with experimental observation that
anthracene is the major product (Yeong, Y. J.; Kang, Y.; et al 2008, Ji, L.; Franke, A.; et al
2014). By contrast, VA shows thermodynamically the dominant hydroxylase as the same as
EC. It seems the precursor leading to aromatic species on desaturation gives the highest
desaturase activity because of the increased resonance stabilization of the benzene product
(Kumar, D.; Tahsini, L.; et al 2009). Regardless of the external factors, we conclude that the
P450-catalyzed desaturation reaction can occur, but is competitive with hydroxylation
reactions on the low-spin state.
To further assess the calculated results, we estimated kinetic isotope effects for the
replacement of the transferring hydrogen atom by deuterium using the semi-classical Eyring
equation as well as the tunneling-corrected Wigner isotope effect (Heyes, D. J.; Sakuma, M.;
et al 2008).
180
Figure 6.4: Potential energy profile of eicosanoic acid activation by 2Cpd I of P450 as
calculated with QM/MM. Energies are given in kcal mol–1
and are calculated at
UB3LYP/BSIV//UB3LYP/BSIII level of theory with ZPE corrections included. Optimized
geometries give bond lengths in angstroms and angles in degrees.
We reevaluated the free energy and imaginary frequency for the hydrogen atom abstraction
step for VA-4,4-d2 and based on a comparison with the fully hydrogenated system establish a
KIE = 6.4 (8.4) using the Eyring model and a KIE = 7.2 (12.1) at the doublet (quartet) spin
states with tunneling corrections included. These values compare well with the
experimentally determined KIE value of 5.58 for the same substrate (Rettie, A. E.; Boberg,
M.; et al 1988).
6.3.3. QM/MM studies of desaturation
As follows from the DFT model calculations in the previous section, both the radical
rebound and second hydrogen atom abstraction processes are barrierless on the doublet spin
2TSH,EA
rCH = 1.333rOH = 1.178
rFeO = 1.730rFeS = 2.420
aFeOC = 123.5
2TSD,EA
181
state surface, whereas a substantial barrier on the quartet spin state surface is found. This
implies that the mechanistic switch from substrate hydroxylation to desaturation by P450-
catalyzed drug metabolism reactions is not kinetically controlled by the properties of the
oxidant and/or substrate but must be affected – at least in part - by external influences, such
as the substrate binding into the substrate binding pocket. Therefore, we followed our study
up by a set of QM/MM calculations on a system that is known to desaturate its natural
substrate. These calculations take the size and shape of the substrate binding pocket into
consideration and should give a realistic picture of the regioselectivity switch from
hydroxylation to desaturation.
The OleTJE P450 isozyme binds long chain fatty acids, such as eicosanoic acid (EA), and
converts them to olefins, CO2 and water using one molecule of oxygen. Figure 6.4 displays
the potential energy landscape as obtained by QM/MM for the hydrogen atom abstraction by
Cpd I from the C–H group and the pathway leading to desaturation. The hydrogen atom
abstraction barrier is low: E+ZPE = 16.4 kcal mol–1
for 2TSH,EA, and of the same order of
magnitude as those reported above for the DFT models as well as to previous hydrogen atom
abstraction barriers from the literature (Kamachi, T.; Yoshizawa, K. 2003, Shaik, S.; Kumar,
D.; et al 2005, Shaik, S.; Kumar, D.; et al 2008, Ogliaro, F.; Harris, N.; et al 2000, de Visser,
S. P.; Kumar, D.; et al 2004, Ji, L.; Schüürmann, G. 2013).
182
Figure 6.5: Geometry scans for the rotation along the Fe–O bond from 2I as calculated
with QM/MM. Energies are given in kcal mol–1
and each data point represents a full
geometry optimization with fixed H–O–Fe–Nheme dihedral angle. Also shown are the maxima
of the scans with key hydrogen bonding interactions identified. The atom labelled with a
yellow star is C.
The hydrogen atom abstraction leads, as above, to an iron(IV)-hydroxo intermediate and
substrate radical. In the case of 2IH,EA the only mechanism we managed to locate is
desaturation to form CO2 and olefin products. Structures of the hydrogen atom abstraction
and desaturation transition states are given in Figure 6.4. The geometry of 2TSH,EA is
analogous to the hydrogen atom abstraction barriers of ethylcarbamate, valproic acid and
ethane, where long C–H and short H–O distances are found to be of 1.33 and 1.18 Å,
respectively. At the same time, the Fe–O bond elongates from about 1.602 Å in 2REA to 1.730
Å in 2TSH,EA, which is indicative of electron rearrangement within the Fe–O bond, see
below.
* = C
** W2 W1
W3
-60
-40
-20
0
20
0 50 100 150 200 250 300
2IH,EA
E
[kcal mol‒1]
dihedral angle H‒O‒Fe‒N [degrees]
11.5
3.8
W2
W1 W3
183
In order to establish pathways leading to desaturation and hydroxylation, we did a series of
geometry scans starting from 2IH,EA, whereby the geometry was stepwise modified along one
degree of freedom, while the rest of the structure was energy minimized. In this particular
case, we searched for pathways from 2IH,EA to either
2IR,EA or
2ID,EA, and as such ran
geometry scans for the dihedral angle rotation (H–O–Fe–Nheme) from 172.6 in 2IH,EA in
gradual steps of 10 in forward and reverse directions for 180. Figure 6.5 gives the obtained
energies for the geometry scans as well as structures of the optimized geometries of the
maxima along the scan in each direction. The clockwise scan (pathway from 2IH,EA to the
right in Figure 6.5) has a small barrier of about 3.8 kcal mol–1
and collapses to alcohol
product complexes by rebound of the OH group to the substrate radical. The structure shows
that this pathway is assisted by hydrogen bonding interactions, whereby the FeOH group
hydrogen bonds to the carboxylic acid group of substrate but also receives a hydrogen bond
from a nearby water molecule in the active site. In this particular P450 isozyme we identified
a large water channel penetrating into the substrate binding pocket and three of these water
molecules are highlighted as W1, W2 and W3 in Figure 6.5. They form clusters of hydrogen
bonding interactions and form a tight network that includes the salt bridge between Arg245
with the carboxylate group of the substrate as well as the channel leading to the propionate
chains of the heme. By contrast, the substrate binding pocket of P450cam is very tight and has
few water molecules present (Poulos, T. L.; Finzel, B. C.; et al 1985, Schlichting, I.;
Berendzen, J.; et al 2000, Auclair, K.; Moënne-Loccoz, P.; et al 2001).
In the anticlockwise direction the scan encounters a much larger barrier of 11.5 kcal mol–1
.
The structure of the maximum of the scan (left-hand-side of Figure 6.5) identifies the key
reason for this high barrier. Thus, upon rotation the OH group of the iron(IV)-hydroxo moiety
clashes with protons of water molecule W1 and hence this pathway is disfavored over the
alternative rotation.
184
Figure 6.6: Valence bond curve crossing diagrams for product formation from radical
intermediates. (a) Radical rebound leading to hydroxylation products. (b) Hydrogen atom
transfer to give desaturation products.
It appears that the subtle network of hydrogen bonding interactions affects the
regioselectivity of the reaction. Note that also this pathway leads to OH rebound and
formation of alcohols. A further geometry scan for the C–COO dissociation in the substrate
leads to release of CO2 and the formation of olefin as shown in Figure 6.4. This pathway has
a lower barrier than the alternative OH rotation of Figure 6.5 and therefore, the desaturation
degree of freedom will be C–C cleavage here rather than OH rotation. The regioselectivity of
hydroxylation versus desaturation, therefore, in OleTJE is guided by the hydrogen bonding
network in the substrate binding pocket that destabilizes the low-spin rebound pathways that
are barrierless in the gas-phase and in model complexes.
6.3.4. Valence Bond rationalization of rebound and desaturation mechanisms
To understand the origin of the regioselectivity of desaturation over hydroxylation we set
up valence bond (VB) curve crossing diagrams that rationalize the kinetics of the reactions.
Figure 6.6 show the VB curve crossing diagrams for the radical rebound and desaturation
185
pathways leading to the competitive hydroxylation and desaturation products. These curve
crossing diagrams start on the left-hand-side with the radical intermediates, i.e. 4,2
IH, which
are described as an iron(IV)-hydroxo (porphyrin) complex with a nearby radical. The wave
function I represents structure 4,2
IH and corresponds to an electronic configuration x2–y22
*xz1 *yz
1 Rad
1 with two unpaired electrons along the Fe–O bond and a third on the substrate
in the Rad orbital. On the bottom right-hand-side of each Figure the product complexes are
drawn, which are either alcohol products or olefin products. The product complexes 4,2
PA and
4,2PD have wave function P and D, respectively, whereby the doublet spin states have
configuration x2–y22 *xz
2 *yz
1 and the quartet spin states x2–y2
2 *xz
1 *yz
1 *z2
1. In a
curve-crossing diagram, as displayed in Figure 6.6, the reactant wave function (I) connects
with an excited state wave function (P* in part a and D* in part b) in the product
geometry. At the same time, the ground state product wave function connects to an excited
state in the reactant geometry (I,reb* in part a and I,D* in part b). These two curves cross
and lead to an avoided crossing and hence a transition state for the reaction pathway from
intermediate to products.
Mathematically, the height of the barrier (E‡) can be determined from a VB curve
crossing diagram by estimating the excitation energy (or promotion gap G) in the reactant
geometry from I to I*. Thus, as has been shown previously (Shaik, S.; Hiberty, P. C.
2007), the barrier height is equal to the curve crossing energy (EX) minus the resonance
energy (B). Since, EX is proportional to a fraction (f) of the promotion gap, the barrier
height can be estimated from:
E‡ = EX – B = fG – B (1)
186
A careful look at the VB structures in the reactant geometry for the hydroxylation pathway
shows that upon excitation the two electrons on the carbon of the substrate and the donor
oxygen atom in I,reb* have formed a bond and at the same time an electron has transferred
from the OH group to iron. The rebound excitation energy Greb, therefore, is dependent on the
strength of the CO bond formed, i.e. the bond dissociation energy of the CO bond (BDECO),
but also on the excitation energy (Eex,FeOH) within the iron-hydroxo complex from 2pz(OH)
into the iron 3d-block, Eq 2. This excitation energy will fill the *xz orbital with a second
electron in the doublet spin state and the *z2 orbital in the quartet spin state.
Greb = BDECO + Eex,FeOH (2)
The desaturation pathway, by contrast, shows somewhat different electron transfer
mechanisms as compared to the hydroxylation pathway, Figure 6.6b. Thus, in the excited
state I,D* the secondary hydrogen atom forms a chemical bond with the hydroxo moiety,
whereas this bond is missing but it forms a bond with the carbon atom of the substrate
instead. Therefore, the excitation energy is proportional to the difference in strength of the O–
H bond that is formed (BDEOH) and the C–H bond that is broken (BDECH). In addition, the
two radicals on the adjacent carbon atoms in I,D* have formed a -bond with energy E.
Obviously, -conjugation will have a major effect on the excitation energy and lower it
dramatically in value. Similarly to the radical rebound also in the desaturation pathway there
is an internal electron transfer from the hydroxo group to the metal (Eex,FeOH), Eq 3.
GD = BDEOH – BDECH + E + Eex,FeOH (3)
In order to stabilize a desaturation pathway over a hydroxylation pathway the barrier for
the latter should be higher in energy than that for the desaturation mechanism, i.e. Ereb‡ >
ED‡. Considering Eq 1 above, the difference in barrier height of radical rebound versus
187
desaturation is then dependent on the promotion gap values G of the two processes and the
resonance energies B. Generally, f is found to have a constant value of 0.30 (Usharani, D.;
Lai, W.; et al 2014). Therefore, if both pathways incur similar resonance energy values B, the
change in product distributions will only be dependent on the relative values of Greb versus
GD, Eq 4.
Greb – GD = BDECO – BDEOH + BDECH – E (4)
The BDEOH value was calculated from the iron(III)-water complex and iron(IV)-hydroxo
intermediate to be 70.8 kcal mol–1
for the iron-porphyrin model used here (Kumar, D.;
Tahsini, L.; et al 2009). The corresponding C–H bond strength of the substrate that is broken
ranges from 36.7 kcal mol–1
for C2H5 to 21.8 kcal mol
–1 for cyclo-C6H7
. By contrast, the C–
O bond strength of the alcohol product that is formed ranges from BDECO = 85.8 kcal mol–1
for ethane to 59.7 kcal mol–1
for cyclohexadiene. Furthermore, DFT calculations predicted a
-conjugation energy of about 36 kcal mol–1
for a benzene molecule (Fernández, I.; Frenking,
G. 2006). Based on this, we calculate a difference Greb – GD = +16 kcal mol–1
for a molecule
like ethane, whereas a value of –25 kcal mol–1
is found for cyclohexadiene. This implies that
a regioselective hydroxylation of ethane may be expected, whereas the desaturation will be
favorable for cyclohexadiene. These values, therefore, implicate strong effects of the nature
of the substrate on the regioselectivity of hydroxylation over desaturation, which is
determined by a much stronger C–O bond formed in ethanol (by 26 kcal mol–1
) than the
corresponding alcohol of cyclohexadiene and in addition also has a stronger C–H bond for
the second hydrogen atom that is abstracted in the desaturation process (by another 15 kcal
mol–1
).
188
As the difference Greb – GD for ethane is calculated to be +16 kcal mol–1
in favor of
substrate hydroxylation, this implies that perturbations through binding of the substrate in the
enzyme binding pocket will have to overcome 16 kcal mol–1
in order to change the
regioselectivity of hydroxylation over desaturation. As seen in Figures 6.4 and 6.5, an
aliphatic substrate like eicosanoic acid can be desaturated by P450 enzymes due to a tight
substrate binding pocket that through hydrogen bonding networks prevent the OH rotation
and, thereby, blocks the OH rebound process efficiently.
Subsequently, we calculated the difference Greb – GD for the drug molecules EC and VA as
well as for DHA and find values of 10, 13 and –4 kcal mol–1
, respectively. This implies that
the dihydroanthracene reaction will naturally be driven towards desaturation and the
formation of anthracene products. However, in the case of EC and VA the balance in the
isolated system is in favor of hydroxylation rather than desaturation. The P450 isozymes
involved in detoxification of EC and VA, therefore, have a specific substrate binding pocket
that will change the balance in favor of desaturation rather than hydroxylation due to
environmental perturbations affecting the rebound process. Consequently, subtle
perturbations in the substrate binding pocket can have a major effect on the regioselectivity of
substrate activation by P450 isozymes and are necessary to change the preference from
hydroxylation to desaturation. This may be an important point to consider into bioengineering
P450s for commercial purposes.
6.3.5. Molecular orbital rationalization of rebound and desaturation mechanisms
To further understand the intricate details of desaturation and hydroxylation reactions by
P450 enzymes, we did a molecular orbital analysis of the relevant orbitals in the reaction
process. The bifurcation originates in the radical intermediate IH, and, therefore, we looked
into the orbital levels of IH as compared to those in PA and PD. The change in orbitals should
reflect the height of the barrier as seen also above in the VB discussion.
189
Thus, the radical intermediate has configuration x2–y22 *xz
1 *yz
1 Rad
1 and upon product
formation the xz2 *xz
1 pair of orbitals splits back into atomic orbitals to form 3dxz,Fe
2 2px,O
1.
Radical rebound via TSreb will lead to the formation of a O–C bond through the mixing of the
2px,O and Rad orbitals leading to the new pair O–C and *O–C, Scheme 6.4. On the other
hand, the desaturation reaction will lead to the mixing of 2px,O, Rad and C–H orbitals of the
radical intermediate into new O–H/*O–H and C=C orbitals. The possibilities and easiness of
this mixing will affect the kinetics of the rebound and desaturation barriers.
Scheme 6.4: Orbital mixing patterns for the pathways from radical intermediates to products.
Therefore, the relative stability of the desaturation versus hydroxylation product complex
will depend on the relative energies of the O–H2 C=C
2 orbitals with respect to the C–H
2 O–C
2
ones. Obviously if the energy of the O–H and C=C orbitals is lower than that of the sum of
the C–H and O–C pair then desaturation is more exothermic.
In a similar way, we could look at the orbitals of the radical intermediate IH. In both
processes the 2px,O and Rad orbitals are involved and are being rehybridized during the
190
pathway leading to products. Along the desaturation, however, also the C–H orbital in IH
comes into play. We, therefore, analyzed the energy gap (E-) between Rad and C–H in the
radical intermediates for several substrates at the B3LYP/6-31+G** level of theory. We find
values of E– of 94.0, 99.7, 103.0 and 132.2 kcal mol–1
for ethylcarbamate, valproic acid,
dehydroanthracene and ethane, respectively. Clearly, the large energy gap between the Rad
and C–H orbital in a molecule like ethane will drive the reaction towards hydroxylation as
indeed also observed experimentally.
In addition to the above mentioned drug molecules, a range of substrates have been
identified as leading to desaturation products from P450 metabolized reactions. In particular,
further examples include the desaturation of testosterone to 17-hydroxy-4,6-androstadiene-
3-one (Nagata, K.; Liberato, D. J.; et al 1986), warfarin to dehydrowarfarin (Kaminsky, L. S.;
Fasco, M. J.; et al 1980), and lindane to 1,2,3,4,5,6-hexachlorocyclohexene (Chadwick, R.
W.; Chuang, L. T.; et al 1975). To test the hypothesis that E– can predict a desaturation
versus hydroxylation product ratio, we then calculated the radical intermediates for
testosterone, warfarin and lindane and find values of 99.7, 106.2 and 106.0 kcal mol–1
,
respectively. These three molecules also give low E– values and are expected to give
desaturation products at least in competition with hydroxylation.
191
6.4 Conclusion
We present here a detailed set of computational results on the potential regioselectivity of
aliphatic hydroxylation versus desaturation. We identify key steps in the mechanisms that
both are initiated by a hydrogen atom abstraction step followed by OH isomerization to form
either the pro-hydroxylation radical intermediate or the pro-desaturation radical intermediate.
A VB and molecular orbital analysis identifies the electron transfer processes for both
reaction steps and highlights the substrate specific factors and orbitals that drive the
regioselectivity to hydroxylation or desaturation. Further evidence from QM/MM
calculations shows that the regioselectivity can be modified through external perturbations
that block one of the isomerization pathways.
192
CHAPTER 7
PROJECT FIVE
193
A trimetal carbene with reactivity Reminiscent of
Fischer-Tropsch Catalysis.5
ABSTRACT
Metal-carbenes are common reaction intermediates in chemical catalysis, however, very few
catalysts are known with a trimetal carbene. In this work we describe the chemical and
catalytic properties of a unique trimetal carbene containing an Ru2Pt-carbene core:
[(CpRu)2(2-CH2)(3-NCH3)Pt(P(CH3)3)2]. The chemical structure and individual orbital
interactions are identified from a natural bond orbital analysis and the reaction pathways for
consecutive intramolecular CH3 and H transfer (or vice versa) are investigated. It is shown
that this trimetal carbene can catalyze alkyl chain growth efficiently through an initial CH3
transfer followed by H migration. These reaction processes are rationalized by valence bond
models and thermochemical bond strength studies. The work leads to suggestions on how to
improve this chemical system for enhanced alkane synthesis.
5 Mala A. Sainna, Devendra Singh, Deve sh Kumar, Sam P. de Visser. “A trimetal carbene with reactivity Reminiscent of Fischer-Tropsch
Catalysis”Organomettalics, 2015, (34), 1651-1660
194
7.1 Introduction
Ruthenium catalysts are very efficient and powerful and are being applied to a range of
chemical processes, including, for instance, olefin metathesis (Fürstner, A. 2000, Monsaert,
S.; Lozano Vila, A.; et al 2009, Vougioukalakis, G. C.; Grubbs, R. H. 2010, Lin, Y. A.;
Davis, B. G. 2010), water oxidation (Hurst, J. K. 2005, Sala, X.; Romero, I.; et al 2009,
Romain, S.; Vigara, L.; et al 2009, Concepcion, J. J.; Jurss, J. W.; et al 2009), the
enantioselective C–C bond formation and oxygen atom transfer reactions (Ikariya, T.;
Murata, K.; et al 2006, Fukuzumi, S.; Ohkubo, K. 2010, Man, W.-L.; Lam, W. W. Y.; et al
2014) . Of these processes, olefin metathesis is a carbon-carbon double bond rearrangement
reaction, where two olefins in a chemical reaction convert to two new olefins (Fürstner, A.
2000, Monsaert, S.; Lozano Vila, A.; et al 2009, Vougioukalakis, G. C.; Grubbs, R. H. 2010,
Lin, Y. A.; Davis, B. G. 2010). This reaction generally happens on an organometallic reaction
center often with Ru as the central catalyst, via the binding of a carbene group to the metal.
The catalytic mechanism of olefin metathesis involves a C–C bond breaking and the
formation of another C–C bond in a subsequent step. In most of these olefin metathesis
catalysts, the metal binds to a carbene either via a single or double bond, whereas carbon
centers ligated to multiple metal atoms are rarely found.
Over the past 30 years quite a number of ruthenium complexes have been identified that are
able to oxidize water to molecular oxygen (Hurst, J. K. 2005, Sala, X.; Romero, I.; et al 2009,
Romain, S.; Vigara, L.; et al 2009, Concepcion, J. J.; Jurss, J. W.; et al 2009, Huynh, M. H.
V.; White, P. S.; et al 2001, Meyer, T. J.; Huynh,M. H. V. 2003, Wang, L.-P.; Wu, Q.; et al
2010, Duan, L.; Bozoglian, F.; et al 2011, Barnett, S. M.; Goldberg, K. I.; et al 2012). These
complexes can either be mononuclear or binuclear ruthenium centers and reduce water via
proton-coupled electron transfer (PCET) to molecular oxygen, for instance, as a means to
generate electricity. In addition, ruthenium catalysts have been investigated extensively in
195
porphyrin chemistry as a replacement of iron. These porphyrin complexes are based on
enzymatic systems including the cytochromes P450 that catalyze substrate hydroxylation and
epoxidation processes on an iron-heme center (Sono, M.; Roach, M. P.; et al 1996, Groves, J.
T. 2003, Ortiz de Montellano, P. R., 2004, Denisov, I. G.; Makris, T. M.; et al 2005, Kadish,
K. M.; Smith, K. M.; et al 2010, O'Reilly, E.; Koehler, V.; et al 2011). Biomimetic studies on
iron-porphyrin complexes investigated the reactivity and chemical properties of the active
site of P450 enzymes and gave insight into the fundamental properties of the catalyst that
drives the reaction. In early studies, Groves and co-workers(Groves, J. T.; Quinn, R. 1985,
Groves, J. T.; Ahn, K.-H. 1987, Groves, J. T.; Ahn, K.-H.; et al 1988, Groves, J. T.; Roman,
J. C. 1995, Groves, J. T.; Bonchio, M.; et al 1996) replaced iron by ruthenium in these
biomimetic porphyrin scaffolds and did a detailed comparison that focused on the reactivity
changes upon metal-replacement. They found ruthenium-porphyrin complexes to react more
efficiently with olefines than the corresponding iron analogues and, moreover, produced
stereospecific reaction products. Subsequent computational studies (Sharma, P. K.; de Visser,
S. P.; et al 2003) showed that this was due to the fact that ruthenium is stabilized in a higher
oxidation state than the corresponding iron porphyrin (Por) complex, i.e. RuV(O)(Por) versus
FeIV
(O)(Por+
), which resulted in differences in electron transfer processes and radical
rebound barriers. In these chemical systems, the active oxidant reacts with aliphatic groups
via proton-coupled electron transfer (PCET), whereby the proton is transferred to the oxygen
atom and the electron to the metal in a formal hydrogen atom abstraction step. PCET is a
common reaction type in transition metal catalysis and many examples have been identified
in the literature over the years (Crevier, T. J.; Mayer, J. M. 1998, Crevier, T. J.; Lovell, S.; et
al 1998, Bryant, J. R.; Mayer, J. M. 2003, Matsuo, T.; Mayer, J. M. 2005, Wu, A.; Mayer, J.
M. 2008, Maestri, A. G.; Cherry, K. S.; et al 2001).
196
Recently, the Matsuzaka group (Takemoto, S.; Morita, H.; et al 2009) reported the
synthesis and spectroscopic characterization of [(Cp*Ru)2(2-CH2)(3-NPh)Pt(P(CH3)3)2],
which was treated with CH3OTf in diethylether and after heating in toluene produced a
unique structure (labelled as 3 in Ref 9) with a trigonal sp2 hybridized carbon atom linked to
three transition metal atoms. It was found that structure 3 reacted via alkyl chain growth via
H and CH3 migration from the RuH and PtCH3 ligands to form 2-CHCH3 products, Scheme
7.1. Therefore, structure 3 catalyzes the C–C and C–H bond formation reactions for the
synthesis of alkanes, which are thermodynamically difficult processes. It may very well be
that the reaction displayed in Scheme 7.1 represents a novel catalytic reaction for the
synthesis of alkanes on carbene centers, but in order to explore its potential and possibilities
further it is important to fully understand the origin of the catalysis and the chemical
properties that drive the reaction mechanisms. Currently, however, there is limited knowledge
on the chemical system displayed in Scheme 7.1 and experimental studies failed to identify
intermediates along the reaction mechanism. Hence it is unclear how and why this catalyst
operates and how it can be further improved. Moreover, there are very few examples in the
literature of analogous processes and this chemical system may pose a novel synthetic route
for the generation of alkanes on a carbene center.
Scheme 7.1: Catalyst investigated in this work for alkyl formation on a carbide center.
197
The aim of the studies in this work, therefore, is focused on establishing the reaction
mechanism of alkyl chain growth on the carbene center and the chemical properties of the
catalyst shown in Scheme 7.1. Firstly, it is unclear whether there is an initial H-transfer
followed by CH3-transfer or the reverse. Moreover, it may very well be that both mechanisms
are kinetically and thermodynamically possible. Secondly, as shown in iron-porphyrin
chemistry for many examples the reaction can take place via proton transfer, hydrogen atom
transfer or hydride transfer (Mayer, J. M. 1998, Weinberg, D. R.; Gagliardi, C. J.; et al 2012).
Furthermore, hydrogen atom and hydride transfer processes may happen via PCET, whereby
the electrons move to a different donor site than the protons involved in the reaction. The
reaction described in Scheme 7.1 shows reminiscence with the ruthenium-porphyrin
reactivity described above, where either hydrogen atom transfer or PCET takes place from
substrate to metal-oxo or metal-nitrido oxidant. Consequently, there are many possibilities for
a chemical reaction here, which warrants a detailed computational investigation. Theory is
the ideal tool to distinguish between these mechanistic pathways and to establish the
electronic features that drive this important mechanism in chemical catalysis. In addition to
the reactivity questions and the synthesis of alkyl groups on this carbon center, there are also
questions related to the chemical structure of the central carbon atom in 3 itself. Thus, an sp2
hybridized carbon atom would make 3 a carbocation with planar conformation, which is
unlikely given the fact that the carbon atom is surrounded by positively charged metal ions.
On the other hand, if the carbon atom is negatively charged, i.e. a carbanion, it would have
pyramidal structure because of its sp3 hybridization, which is in disagreement with the
experimental crystal structure coordinates. Clearly, the structural and reactivity features of
this complex are poorly understood and theory may provide insight into the electronic and
structural features of the catalyst. Moreover, as it appears this is a novel catalytic system that
may be exploited further for the synthesis of linear alkanes from smaller components, i.e.
198
methane and hydrogen, it may serve as a template for the design and development of future
catalysts.
The Ru2PtC core of structure 3 is an unusual structure in chemical catalysis but shows
some resemblance to Grubbs catalyst (Love, J. A.; Sanford, M. S.; et al 2003, Alcaide, B.;
Almendros, P.; et al 2009, Bernal, M. J.; Torres, O.; et al 2013), where the metal is covalently
linked to a carbon atom through either a single, a double or a triple bond. However, as far as
we know, there are no reported catalysts in the literature, where the carbon atom is ligated to
two Ru atoms and a Pt atom. There are several examples in the literature of carbido centered
structures, where the carbon atom is bound to three or more metal atoms (Miller, R. L.;
Wolczanski, P. T.; et al 1993, Su, C.-J.; Su, P.-C.; et al 1996, Peters, J. C.; Odom, A. L.;
1997, Caselli, A.; Solari, E.; et al 2000, Buchowicz, W.; Herbaczyńska, B.; et al 2012,
Harding, D. J.; Kerpal, C.; et al 2013, Borren, E. S.; Hill, A. F.; et al 2013), but very few of
these show catalytic activity. It, therefore, remains unclear why structure 3 is catalytically
active for the formation of alkyl groups from individual methyl, hydrogen and carbon groups.
In order to gain insight into the peculiar structure and catalytic reactivity of 3 we decided to
do a computational study and establish the electronic and thermochemical features. As we
will show here our studies give fundamental insight into the nature of the metal-carbon bond
and characterizes structure 3 as a trimetal carbene that can react via sequential methyl and
hydrogen atom transfer to form 2-CHCH3. These results may have direct relevance to
heterogeneous catalysis on metal surfaces.
7.2 Methods
The calculations reported in this work were performed using density functional theory (DFT)
methods as used previously on transition metal complexes by our groups (de Visser, S. P.
2010). We used DFT methods as implemented in the Jaguar 7.7 (Schrodinger, LLC, 2011)
199
and Gaussian-09 (Frisch, M. J.; et al 2009) program packages. Initial geometries were
optimized (without constraints) in Jaguar using hybrid density functional theory method
(B3LYP) (Becke, A. D. 1993, Lee, C.; Yang, W.; et al 1988) coupled with an LACVP basis
set on ruthenium and platinum and 6-31G on the rest of the atoms: basis set BS1 (Hay, P. J.;
Wadt, W. R. 1985). A subsequent analytical frequency in Gaussian confirmed the structures
as local minima with real frequencies only or first order saddle points with one imaginary
frequency for the correct mode. Energies were then improved by running a single point
calculation in Jaguar using a triple- quality LACV3P+ basis set on iron and 6-311+G* on
the rest of the atoms: basis set BS2. We also tested the effect of dispersion on the reaction
energies by doing a UB3LYP-D3 single point calculation in Jaguar using the model of
Grimme and co-workers (Grimme, S.; Antony, J.; et al 2010). For a selection of chemical
systems, i.e. reactants and transition states, we did geometry optimizations at UB3LYP/BS2
and UB3LYP-D3/BS2, but found very little changes with respect to those obtained at
UB3LYP/BS1. Since, transition metal complexes often react via multi-state-reactivity (Shaik,
S.; de Visser, S. P.; et al 2002), we calculated the complete potential energy profile on the
lowest lying singlet, triplet and quintet spin states, and however, in all cases the singlet spin
state was the ground state and well separated from the other spin states.
As occasionally DFT calculations on spin state ordering and relative energies of transition
metal complexes shows fluctuations depending on the choice of the density functional
method (Quesne, M. G.; Latifi, R.; et al 2014), we decided to do a series of test calculations
with alternative DFT methods, namely B3LYP-D3 (Grimme, S.; Antony, J.; et al 2010),
BP86 (Becke, A. D. 1988, Perdew, J. P.; Burke, K.; et al 1996), MO6 (Zhao, Y.; Truhlar, D.
G. 2008), and PBE0 (Adamo, C.; Barone, V. 1999, Perdew, J. P.; Burke, K.; et al 1996).
These studies confirmed the trends and did not change the conclusions.
200
The methods described here were used previously for analogous transition metal containing
systems in comparison with experimentally determined data and reproduced free energies of
activation within 4 kcal mol–1
(Kumar, D.; de Visser, S. P.; et al 2005, Vardhaman, A. K.;
Sastri, C. V.; et al 2011), but also gave reasonable agreement for spectroscopic data
(Karamzadeh, B.; Kumar, D.; et al 2010). To locate transition state structures we ran
extensive geometry scans between the various local minima, and used the maximum of those
scans as a starting point for the actual transition state searches. Moreover, these scans
confirmed that reactants and products indeed connect via the transition state.
Our chemical model is based on the crystal structure coordinates of [(Cp*Ru)2(2-H)(2-
NHPh)(3-C)PtCH3(P(CH3)3)2]+ (Takemoto, S.; Morita, H.; et al 2009) whereby we replaced
the Cp* groups with Cp and phenyl by methyl. The stereochemical effects of the methyl
groups of the P(CH3)3 groups on the reaction mechanism was investigated by replacement of
the P(CH3)3 groups with PH3. The latter structures are identified as 3’, IMe’ and IH’.
Charges reported in this work were obtained from a Natural Bond Occupation (NBO)
calculation as implemented in Gaussian program package at B3LYP/BS2 level of theory
(Weinhold, F.; Carpenter, J. E. 1988).
In order to fully understand the thermodynamics and kinetics of the mechanism of the
overall chemical reaction, we calculated the bond dissociation free energies (BDFE) of key
Ru/C–H and Pt/C–CH3 bonds in complexes 3, IMe, IH and PH. The release of either H•, H
+,
CH3• and CH3
+ from these four complexes as described by Eqs 1 – 4, whereby each
individual structure was calculated through a full geometry optimization to give the adiabatic
bond dissociation free energies. In addition, we calculated the methylenium transfer free
energy (MTFE) and the proton transfer free energy (PTFE). The latter was calculated through
the use of H2O/H3O+ couple to balance the equation. As there were large structural changes
201
upon H/CH3 bond breaking on the various complexes we also calculated the diabatic bond
dissociation free energies, or vertical bond dissociation free energies, whereby we took the
optimized geometries of 3, IMe, IH and PH and ran single point calculations with either H•, H
+,
CH3• or CH3
+ removed.
3+ [3 – H]
+ + H
• + BDFEH (1)
3+ [3 – CH3]
+ + CH3
• + BDFEMe (2)
3+ + H2O [3 – H]
0 + H3O
+ + PTFE (3)
3+ [3 – CH3]
0 + CH3
+ + MTFE (4)
7.3 Results and Discussion
Before we discuss our calculated reaction mechanisms, let us briefly summarize the main
results on our reactant structure, 3. The optimized geometry of 3 is shown in Figure 6.1 and
gives bond distances to within 0.011 Å of the crystal structure coordinates reported by
Takemoto and co (Takemoto, S.; Morita, H.; et al 2009) and as such the DFT calculations
give reasonable chemical structures. The Pt–C bond of 1.960 Å represents a single bond and
is in good agreement with several reported Pt–C containing crystal structures from the
literature (de Quadras, L.; Bauer, E. B.; et al 2007, Che, C.-M.; Huang, J.-S. 2002, Das, R.
K.; Saha, B.; et al 2010, Algarra, A. G.; Grushin, V. V.; et al 2012). The two Ru–C bonds are
very close in bond length and the fact that one of the two atoms is protonated, therefore,
appears to have little effect on its interaction with carbon. Both Ru–C bonds have a typical
length corresponding to a single bond and compare to analogous structures reported before
202
(Mutter, S. T.; Platts, J. A. 2011, Conner, D.; Jayaprakash, K. N.; et al 2003, Patra, S. K.;
Sadhukhan, N.; et al 2006, Patra, S. K.; Bera, J. K. 2006, MacGregor, S. A.; McKay, D.; et al
2013). As transition metal complexes containing iron or ruthenium generally contain close-
lying spin states and often react via multistate reactivity patterns (Shaik, S.; de Visser, S. P.;
et al 2002, Latifi, R.; Sainna, M. A.; et al 2013), structure 3 was optimized in the singlet,
triplet and quintet spin states. The closed-shell singlet spin state is the ground state and is
well separated from the triplet and quintet spin states by Ggas = 17.1 and 44.8 kcal mol–1
,
respectively. These values change by less than 1 kcal mol–1
when solvent corrections are
included. Clearly, catalyst 3 is a closed-shell species that will react via single-state reactivity
in contrast to metal(IV)-oxo and metal(V)-nitrido complexes from the literature (Vardhaman,
A. K.; Barman, P.; et al 2013).
Figure 7.1: Optimized geometry of 3 as calculated with B3LYP. Bond lengths are given in
angstroms and group NBO charges Q in atomic units. The right-hand-side displays the
natural bond orbitals and their ordering for those involving the central carbon atom with it
ligands.
1.9701.960
1.938
rRu-Ru = 2.531rRu-H = 1.610rPt-Me = 2.119
QC = 0.04QRu = –0.05QRuH = –0.19QPtMe = –0.32
203
In order to shed light on the bonding pattern of the central carbon atom, we did a Natural
Bond Orbital (NBO) analysis of structure 3. Thus, the NBO analysis reveals three bonding
type orbitals for the central carbon atom, namely three single bonds for the interaction with
the two Ru atoms and the Pt atom. The PtC orbital represents the -bonding orbital between
Pt and C and is built up from the interaction of the 5dz2 orbital on Pt with the 2pz orbital on C.
In addition, there are two single bonds between the carbido atom and the two Ru atoms,
which result from the mixing of the 2s, 2px and 2pz atomic orbitals on C with the 4dz2 orbital
on Ru or the 4dxz on Ru(H). As such the central carbon atom is sp2 hybridized. However, the
fourth valence orbital on carbon is a virtual 2py atomic orbital. The NBO analysis, therefore,
assigns the central carbon atom as a carbene with three sp2 hybridized orbitals that are
occupied with two electrons in an orbital donating to Pt (PtC), and two bonding orbitals
where both C and Ru donate one electron each into the bond (RuC, RuC). This hybridization
scheme gives the central carbon atom its planar structure, Scheme 7.2. As such the chemical
structure of Ru2CPt represents a Fischer-carbene, where the lone pair of the carbon atom
interacts with an empty Pt orbital and the -type Pt orbitals give backbonding into the empty
2py orbital of the carbine (Vardhaman, A. K.; Barman, P.; et al 2013). The NBO charges
displayed in Figure 7.1 reveal a charge-neutral carbon atom (QC = 0.04) that is surrounded by
slightly negatively charged metal atoms. Consequently, there is very little charge built-up in
this chemical system.
Scheme 7.2: Hybridization scheme of 3.
204
Subsequently, we calculated the potential energy profile for 2-CHCH3 formation starting
from 3. We tested two mechanisms as shown in Figure 7.2: (i) initial hydrogen transfer from
RuH to the carbene followed by methyl abstraction and (ii) initial methyl transfer from PtCH3
followed by hydrogen/proton transfer. Figure 7.2 starts from the center with 3 and follows the
two possible reaction mechanisms: The initial H-transfer (mechanism to the right) and the
initial CH3 transfer (mechanism to the left). The two mechanisms are described as follows: In
the first case, the metal-carbene group abstracts a hydrogen atom from the RuH group via a
transition state TSH to form the H-transfer intermediate IH. A subsequent CH3 transfer from
the PtCH3 group via a barrier TSreb,Me gives the 2-CHCH3 bound product PH. We also tested
the alternative pathway starting with an initial CH3 abstraction by the metal-carbene group
via transition state TSMe leading to the methyl-transfer intermediate IMe. This intermediate
then reacts further by H-abstraction from RuH via a transition state TSreb,H to form product
PMe.
Figure 7.2: Free energy profile of alkyl chain growth on 3 via either (i) CH3-transfer
followed by H-transfer (mechanism from the center to the left) or (ii) H-transfer followed by
CH3-transfer (mechanism from the center to the right). Free energies (in kcal mol–1
) are
TSHTSMe
rPtC = 1.987rRuH = 1.596
rHC = 1.446rRuH = 1.738
rRuC = 1.938rPtCH3 = 2.113i629.7 cm–1
rPtCH3 = 2.355rCC = 1.824i352.4 cm–1
IMe
rPtRu = 2.820rPtRu = 2.957rRuRu = 2.548dCRuRuPt = 50.4
IH
rPtRu = 3.086rPtRu = 3.370rRuRu = 2.620dCRuRuPt = 43.3
205
obtained with B3LYP-D3/BS2 and contain zero-point, thermal and entropic corrections at
298K. Values in parenthesis include solvent corrections to the free energy. Optimized
geometries report bond lengths in angstroms, angles in degrees and the imaginary frequency
in the transition states in wave numbers.
The H-atom transfer reaction has the lowest free energy of activation of Ggas (Gsolv) =
17.3 (17.6) kcal mol–1
, however, its subsequent CH3-transfer to form products gives an
additional barrier with a maximum of 43.0 (42.4) kcal mol–1
. Therefore, the reaction that is
initiated with H-atom transfer has an unfeasible free energy of activation for TSreb,Me and will
not be able to proceed. This is in agreement with experimental observation that found
evidence of a structure with analogy to IH, but the methyl group bound to one of the Ru
atoms. Clearly, any IH formed during the reaction is a dead-end product and will have to react
back to reactants to be catalytically active.
The alternative pathway with an initial CH3-transfer, on the other hand, has a calculated
free energy of activation of Ggas (Gsolv) = 21.4 (20.9) kcal mol–1
, and, therefore, is
disfavored over the initial H-atom transfer process. However, this reaction leads to a very
stable intermediate 1IMe with an exergonicity of almost 10 kcal mol
–1. Its subsequent H-atom
transfer leading to products only has a free energy of activation of 19.1 (19.0) kcal mol–1
from intermediates and therefore product formation is a viable pathway through this channel
with a rate determining TSMe barrier. Consequently, DFT establishes a possible alkyl
synthesis mechanism for 3 as a stepwise mechanism with an initial CH3-transfer followed by
H-atom transfer. An analysis of the group spin densities and charges along the complete
reaction mechanism shows that all local minima and transition states are closed-shell singlet
species with no radical built-up during the reaction. Accordingly, the transfer of a H-atom or
206
CH3-group does not leave a radical center behind but instead the charges are redistributed
over the chemical system to retain maximum electron pairing.
To test whether changing the density functional method would have an effect on the free
energy of activation in the chemical reactions, we decided to calculate the energies of 3, TSMe
and TSH using B3LYP-D3, B3LYP, BP86, PBE0 and MO6 on the optimized geometries of
Figures 7.1 and 7.2. Figure 7.3 displays the energy splitting of 3, TSMe and TSH as calculated
with a range of density functional theory methods. As follows from the results in Figure 7.3,
the ordering of the barrier heights stays the same between all DFT methods and the absolute
values of the barrier heights change by less than 3.5 kcal mol–1
maximally. Moreover, the
energy splitting between TSH and TSMe is calculated to be between 4.6 kcal mol–1
(for BP86)
and 7.1 kcal mol–1
(for B3LYP). Clearly, as the singlet spin ground state is well separated
from the higher spin states, the effect of amount of exchange in the DFT calculation has little
effect on the thermodynamics and kinetics of the reaction and all methods provide the same
qualitative result. An analysis of group spin densities and charges of these structures
calculated with alternative DFT methods gave no changes and confirmed the conclusions
drawn above.
The calculations presented in this work show that the reaction starts with an H/CH3 transfer
and no radical center is left behind. We made several attempts to swap molecular orbitals and
find alternative intermediates and transition states, corresponding to H+/CH3
+ transfer
intermediates or H–/CH3
– transfer intermediates but in all cases the structures converged back
to the lower-lying H-atom/CH3-transfer pathways or led to higher energy conformers.
Clearly, the cationic and anionic pathways are considerably higher in energy and will not be
relevant for the reaction mechanism, which we will rationalize with molecular orbital
analysis and thermochemical cycles below.
207
Figure 7.3: Energies of initial H-atom or CH3-transfer reactions from 3 as calculated with
different DFT methods. All structures optimized at B3LYP/BS1 and single point calculations
with basis set BS2 applied. Relative energies are given in kcal mol–1
and include ZPE
corrections calculated at B3LYP/BS1.
So why is the reaction displayed in Figure 7.2 stepwise with consecutive CH3 and H-
transfer processes and what is the function of the Ru2Pt scaffold? These questions will be
answered in the following discussion. Structurally, there are major changes happening at the
Ru2Pt-carbene center during the reaction mechanism. Although 3 has an sp2 hybridized
carbon atom that lies in the plane of the Ru2Pt atoms, upon H/CH3-transfer the planarity is
distorted and the carbon atom relaxes to a pyramidal conformation due to sp3 hybridization.
The pyramidalization is considerably more pronounced for the CH3-transfer process than in
the H-atom transfer pathway as dihedral angles C–Ru–Ru–Pt of 50.4 for IMe and 43.3 for IH
are found. At the same time the Pt–Ru distances in IMe shorten by as much as 0.8Å with
208
respect to 3 to well below 3Å, while in IH they stay considerably longer: rPtRu = 3.086 and
3.370Å. These large structural changes for the CH3-transfer pathway will have an effect on
the reaction energetics and kinetics as well as on the relative stability of IMe versus IH. Thus,
the Pt atom in IMe only has three ligands due to CH3-transfer to the carbon atom, which
means limited stereochemical interactions between the two P(CH3)3 groups and the rest of the
molecule so that the Pt atom can approach the two Ru atoms closely. In IH the Pt atom
remains tetracoordinated and the stereochemical interactions between all ligands and the rest
of the molecule prevent a close approach to the Ru2C center. As a consequence of the
stereochemical repulsions IH is considerably destabilized over IMe and the reaction pathway
with initial H-atom transfer is energetically disfavored.
To further establish these stereochemical interactions in IH versus IMe, we did a subsequent
set of calculations on these structures, but with the P(CH3)3 groups replaced by PH3:
structures IH’ and IMe’, Figure 7.4. The optimized geometry of 3’ (Figure 7.4) is very close to
that reported above for 3 in Figure 7.1 and only minor changes in bond lengths are obtained.
Clearly, the P(CH3)3 groups have little effect on the structure of the Ru2PtC core of our
chemical system. A similar situation occurs for IMe’, which has large structural similarities
with IMe reported in Figure 7.2. Major changes, however, are seen when IH is compared to
IH’. Thus, one of the Ru–Pt distances decreases from 3.370 to 3.116 Å and as a consequence
the pyramidality of the carbon center changes from dCRuRuPt = 43.3 in IH to a value of 51.9
in IH’. This has a major effect on the stability of IH’ and it is stabilized by 6.0 kcal mol–1
with
respect to 3’, however, the IMe’ remains the most stable conformation. This implies that the
P(CH3)3 groups incur a destabilizing effect on the reaction mechanism, but will not reverse
the preference of initial CH3-transfer over H-atom transfer.
209
Figure 7.4: Optimized geometries of 3’, IMe’ and IH’ with bond lengths in angstroms.
To gain insight into the electronic changes during the reaction mechanism, a NBO analysis
was also performed on the intermediate and product complexes. The NBO calculation reveals
a configuration on the central carbon atom of 2s1.08
2p2.92
for IMe and 2s1.18
2p3.10
for IH. Both
intermediate complexes, therefore, have an electronic configuration close to sp3
hybridization, and as a consequence their geometries have changed from planar to pyramidal
in agreement with the optimized structures.
In order to fully understand the thermodynamics and kinetics of the mechanism displayed
in Figure 7.2, and, in particular, gain insight into the electron transfer mechanisms during the
reaction processes we set up Valence Bond (VB) curve crossing diagrams (Figure 7.5) for the
methyl transfer and hydrogen atom transfer steps. The VB curve crossing diagrams start at
the bottom-left with the reactant electronic configuration and wave function (3) of 3, which
is a carbene where two sp2 electrons form bonding orbitals with electrons on the two
ruthenium atoms and the other two electrons represent a lone pair orbital that donates a bond
IH’
1.9561.9651.500
1.935
rRu-Ru = 2.534rRu-H = 1.609rPt-C = 2.108
1.089
IMe’3’
rRu-Ru = 2.553rPt-Ru = 2.778rPt-Ru = 2.915rRu-H = 1.604dCRuRuPt = 48.5
rRu-Ru = 2.628rPt-Ru = 3.057rPt-Ru = 3.116rPt-C = 2.122dCRuRuPt = 51.9
210
to Pt. Each of the dots in Figure 7.5 represents one of these valence electrons. In part (a) in
Figure 7.5 the reaction follows the methyl transfer from Pt to carbene and the structure on the
right-hand-side gives the electronic configuration of IMe with wave function Me. In IMe the
central carbon atom is sp3 hybridized and forms a single bond with Ru, Ru, Pt and CH3. In
VB theory the wave functions of the reactant and product complexes cross, i.e. 3 and Me in
Figure 7.5a, and lead to an avoided crossing and a transition state for the reaction from
reactants to intermediates (Shaik, S.; Hiberty, P. C. 2007). Thus, the curve crossing energy
(EX) is above the actual transition state (EMe‡) by a factor B, which is called the resonance
energy. It has been shown that the barrier height can be described by the difference in energy
by a fraction (f) of the promotion gap (G) and the resonance energy B: EMe‡ = fGMT – B. The
promotion gap signifies the excitation energy from the ground state wave function to the
product wave function in the geometry of the reactants, i.e. GMT is the energy difference
between 3 and 3* in the reactant geometry. The VB diagrams also give chemical
structures of the reactant and product wave functions of the ground and excited state species
with valence electrons identified with a dot.
211
Figure 7.5: Valence bond curve crossing diagrams for methyl transfer (part a) and hydrogen
atom transfer (part b) from 3. Valence electrons are identified with a dot.
A close inspection of the VB diagrams in Figure 7.5 shows that the promotion gap for
methyl transfer (GMT) represents the breaking of the Pt–CH3 bond and the formation of a new
C–CH3 bond. In addition, the VB structures of 3 and 3* shows that the carbene carbon is
rehybridized from sp2 to sp
3 hybridization and one electron of the lone pair orbital pointing
toward Pt is promoted into the new sp3 orbital. At the same time the electron from the broken
Pt–CH3 bond moves into the carbene-Pt bond. Thus, the promotion gap also contains a
component for the excitation energy of the carbene from sp2 to sp
3 hybridization (Eex,C),
which essentially requires a promotion of an electron from the PtC orbital to the virtual 2py
molecular orbital on carbon. Energetically GMT can be described in terms of the bond
212
dissociation energies (BDEs) of the respective bonds that are broken and formed in the
reaction process, Eq 5, whereby BDEPtCH3 represents the free energy to break the Pt–CH3
bond in 3 and BDECCH3 is the free energy to form the C–CH3 bond in IMe.
GMT BDEPtCH3 – BDECCH3 + Eex,C (5)
The situation is dramatically different for hydrogen atom transfer from 3 as shown in
Figure 7.5b. Thus, the barrier height for hydrogen atom transfer (EH‡) is proportional to the
promotion gap for hydrogen atom transfer (GHT), which correlates with the strength of the
Ru–H bond that is broken (BDERuH), the strength of the C–H bond that is formed (BDECH)
and again the excitation energy for the carbene to change from sp2 to sp
3 hybridization. In
addition to these three terms, the promotion gap GHT also contains a component for the
electron transfer energy from Ru to Pt (ETRuPt). The hydrogen atom transfer process,
therefore, incurs extra electron reorganization energy in the Ru2Pt scaffold, which is not
necessary for the methyl transfer step. The overall preference of hydrogen atom transfer
versus methyl transfer will depend on the relative strengths of the bonds that are broken and
formed in the process and the electron reorganization energy. To find out whether there are
major differences between the BDE values in Eq 5 and those in Eq 6, we decided to calculate
their values with DFT.
GHT BDERuH – BDECH + Eex,C + ETRuPt (6)
As the VB drawings in Figure 7.5 only give quantitative analysis of electron transfer
processes and no actual qualitative values, we decided to calculate BDEs and bond
213
dissociation free energies (BDFEs) of key Ru/C–H and Pt/C–CH3 bonds in complexes 3, IMe,
IH and PH by calculating the release of a H•, H
+, CH3
• and CH3
+ from these four complexes
according to Eqs 1 – 4 above. Figure 7.6 displays the adiabatic as well as diabatic driving
forces for H/CH3 release from complexes 3, IH and IMe. Thus, in the adiabatic BDFE
calculations we did a full geometry optimization of all individual chemical structures in Eqs 1
– 4 and then took the difference in free energy for each of these reactions. However, due to
considerable rehybridization of the central carbon atom and as a consequence large
geometrical changes, the adiabatic BDFE values gave odd results. For instance, the adiabatic
bond dissociation energy for the breaking of the Ru–H bond in 3 appears negative, which
would imply spontaneous dissociation in disagreement with the experimental crystal
structure. Conversely, when we calculate the diabatic free energy change by taking the H and
[3 – H] products in the geometry of 3 then the BDFE is positive. The same is found for the
dissociation of a methyl group from 3. By contrast to structure 3, the adiabatic and diabatic
bond energies in structures IMe and IH are very similar and vary by less than 6 kcal mol–1
. In
the following, however, we will focus on the diabatic bond dissociation free energies.
Figure 7.6: Bond dissociation free energies (BDFEs in kcal mol–1
) of key bonds in structures
3, IMe and IH. Reactions calculated according to Eqs 1 – 4. Part (a) gives adiabatic BDFE
214
values and part (b) diabatic BDFE values. Values in parenthesis are solvent corrected free
energies, whereas those out of parenthesis are gas-phase data.
Thermodynamically, the driving force to form IMe/IH from 3 represents the breaking of a
weak Pt–CH3 or Ru–H bond and the formation of a much stronger C–CH3/C–H bond. Hence
the reaction should be strongly exergonic, which is seen for the reaction via IMe in Figure 7.2,
but not for the one via IH. Thus, the hydrogen atom transfer from Ru in 3 to the carbene to
form IH results in the breaking of the Ru–H bond of BDFEH,diabatic = 65.7 kcal mol–1
in the
gas-phase and the formation of a C–H bond with a diabatic value of 108.6 kcal mol–1
and, if
the driving force was solely dependent on the change in bond strength the reaction would
have been exergonic by 42.9 kcal mol–1
. As shown in Figure 7.2, however, this reaction is
endergonic by 17.2 kcal mol–1
, which implies that the stereochemical and electronic effects,
i.e. the sp2 to sp
3 rehybridization (Eex,C) and the Ru PT electron transfer (ETRuPt)
accounts for a destabilization of IH by about 60 kcal mol–1
in free energy. This is explained
schematically in Figure 7.7, where we deconvolute the processes from 3 to IMe and from 3 to
IH into factors for bond breaking and formation (BDFE) and quantum mechanical effects
(EQM). Thus, the methyl transfer reaction from 3 results in the breaking of the Pt–CH3 bond,
which requires a BDFEMe,diabatic = 47.6 kcal mol–1
, and at the same time the formation of the
C–CH3 bond (BDFEMe,diabatic = 75.9 kcal mol–1
). Therefore, the change in bond dissociation
free energy for the methyl transfer is –28.3 kcal mol–1
. The reaction mechanism in Figure 7.2
found the methyl transfer process to be exergonic by 9.9 kcal mol–1
, which implies that the
quantum mechanical effect of the process accounts for 18.4 kcal mol–1
. The quantum
mechanical effect includes the rehybridization energy of the carbene group and the geometric
and stereochemical factors associated with the methyl transfer. By comparison, the hydrogen
215
atom transfer shows a much larger change in bond dissociation energy due to the strong C–H
bond that is formed (BDFEH,diabatic = –42.9 kcal mol–1
). However, this reaction incurs a large
quantum mechanical effect of 60.1 kcal mol–1
and thereby makes the overall process highly
endergonic. The quantum mechanical effect includes the rehybridization of the carbon atom,
which will probably of the same order of magnitude as that found for the methyl transfer
process, i.e. about 18 kcal mol–1
. In addition, there are the above mentioned stereochemical
repulsions of the methyl groups of the P(CH3)3 moieties that raise IH by about 6 kcal mol–1
.
Finally, the electronic effects (EQM) for the hydrogen atom transfer reaction include the
electron transfer energy from Ru to Pt as described in the VB diagram in Figure 7.5.
Figure 7.7: Energy decomposition of the methyl and hydrogen atom transfer reaction from 3.
Free energies given are in kcal mol–1
.
In addition to removal of H/CH3 from our reactant complexes and intermediates, the
breaking of the bond into two ions was also investigated and is described as the proton
transfer free energy (PTFE) and methyl transfer free energy (MTFE). As follows from the
data in Figure 7.6, it requires considerably more energy to break the Ru–H and Pt–CH3 bonds
216
than a homogeneous splitting into Ru+H
and Pt
+CH3
. The thermodynamic analysis
confirms the results described above in Figure 7.2 where either an H or CH3
is transferred
rather than H+ and CH3
+.
So what about the next step for the formation of 2-CHCH3 products? Scheme 7.3 displays
the electron transfer processes from IMe via TSreb,H to form products (part a) and those from
IH via TSreb,Me to form products (part b). The reaction from IMe to products results in the
breaking of the Ru–H bond and the formation of a new C–H bond. At the same time the C–Pt
bond breaks and Pt forms a new bond with Ru. By contrast, the pathway from IH leads to the
breaking of the Pt–CH3 bond and the transfer of the methyl group to carbon. However, this
process breaks the C–Pt bond and the Pt(P(CH3)3)2 group is left behind with two unpaired
electrons. Thus, an electron on Pt is promoted into a virtual orbital and the reaction proceeds
via a high energy pathway. Indeed, the DFT calculated potential energy profile in Figure 7.2
gives a high barrier for TSreb,Me in agreement with this. The VB mechanism of Scheme 7.3,
therefore, gives a rationale why the first group migration needs to come from the Pt ligand
and the electronic problems that are obtained when its ligand transfers last.
Scheme 7.3: VB description of second reaction steps. (a) hydrogen-atom abstraction. (b)
methyl-transfer.
217
In summary, the DFT calculations presented here show a novel pathway for the synthesis
of linear alkanes on a trimetal-carbene center. The chemical system discussed here, however,
is inefficient as the lowest lying initial barrier is for hydrogen atom transfer, which leads to a
dead-end reaction intermediate. By contrast, the initial methyl transfer can be followed by
hydrogen atom transfer to lead to products. The question is how this catalyst may be
improved for alkane synthesis. The thermodynamic and VB Schemes above give some clear
indication to this. As shown in Eqs 5 and 6 above, the barrier heights TSH and TSMe are
dependent on the Ru–H and Pt–CH3 bonds that are broken and the C–H and C–CH3 bonds
being formed. One way to raise the hydrogen atom abstraction barrier is by replacing
hydrogen by deuterium. We tested this and find a Ggas = 17.7 kcal mol–1
for transfer of an
Ru–D group to form a C–D bond. However, this is only a minor increase with respect to the
Ggas = 17.3 kcal mol–1
found for hydrogen. An alternative solution to raise the initial
hydrogen atom abstraction barrier is by replacement of the Ru atoms by Os. Bond strengths
increase down the periodic table and it is expected that the BDFEOs–H is larger than the
BDFERu–H, which in itself is above the value of the BDFEFe–H. Indeed, the tabulated bond
strength of an Fe–H diatomic molecule is 43 kcal mol–1
, whereas a Ru–H diatomic molecule
has a reported bond strength of 56 kcal mol–1
(Lide, D. R. 1996) It may very well be that
structure 3 with both ruthenium atoms replaced by osmium is a powerful alkane synthesizing
catalyst that operates through an initial methyl transfer followed by hydrogen abstraction, but
further research will need to establish this.
The work described here may have relevance to biological and heterogeneous catalysis for
the synthesis of alkanes. In particular, in heterogeneous catalysis linear alkanes are generated
on a metal surface using often CO and H2 are reactants. It may very well be that the trimetal-
carbene structure discussed here is a common type intermediate in heterogeneous catalysis,
but further research will have to be performed to establish this.
218
7.4 Conclusions
In summary, our calculations establish the key features of a synthetic homogeneous catalyst
for alkyl chain growth. A combination of DFT, NBO, VB and thermochemical studies has
been performed and established the intrinsic chemical properties of 3. Firstly, we characterize
3 as having a Ru2Pt-carbene core where the lone-pair is donated to Pt via
bonding/backbonding configuration. Secondly, we find a low energy mechanism of alkyl
formation through consecutive CH3 and H
transfer to the carbene with low free energy of
activation and exergonic driving force. We show that the reverse process is thermochemically
hampered and would lead to an excited triplet configuration.
Overall, our work identifies a novel catalyst for the synthesis of alkanes that starts from a
trimetal carbene. This unique structure is shown to be capable of intramolecular methyl and
hydrogen atom transfer to the carbene to form 2-CHCH3 products as precursor to alkanes.
219
CHAPTER 8
PROJECT SIX
220
Does hydrogen bonding-donation to
manganese(IV)-oxo and iron(IV)-oxo oxidants
affect the oxygen atom transfer ability? A
computational study.6
ABSTRACT
Iron(IV)-oxo intermediates are involved in oxidations catalyzed by heme and non-heme iron
enzymes, including the cytochromes P450. At the distal site of the heme in P450 Compound I
(FeIV
-oxo bound to porphyrin radical), the oxo group is involved in several hydrogen bonding
interactions with the protein, but their role in catalysis is currently unknown. In this work we
investigate the effects of hydrogen bonding on the reactivity of high-valent metal-oxo moiety
in a biomimetic model complex that has three hydrogen bond donors directed toward a
metal(IV)-oxo group. We show these interactions lower the oxidative power of the oxidant in
reactions with dehydroanthracene and cyclohexadiene dramatically as they decrease the
strength of the O–H bond (BDEOH) in the resulting metal(III)-hydroxo complex. Furthermore,
the distal hydrogen bonding effect cause stereochemical repulsions with the approaching
substrate and forces a sideways attack rather than a more favourable attack from the top. The
calculations, therefore, give important new insights into distal hydrogen bonding, and show
that in biomimetic, and, by extension, enzymatic systems, the hydrogen bond may be
important for proton relay mechanisms involved in the formation of the metal-oxo
6 Reza Latifi, Mala A. Sainna, Elena Rybak-Akimova,Sam P. de Visser. “Does hydrogen bonding-donation to manganese(IV)-oxo and
iron(IV)-oxo oxidants affect the oxygen atom transfer ability? A computational study” Chemistry A- European Journal. 2013, 19, 4058-4068.
221
intermediates, but the enzyme pays the price for this by reduced hydrogen atom abstraction
ability of the intermediate. Indeed, in nonheme iron enzymes, where no proton relay takes
place, there generally is no donating hydrogen bond to the iron(IV)-oxo moiety.
8.1 Introduction
Hydrogen bonding is widespread around the active sites of oxidative metalloenzymes and has
functions ranging from protein/enzyme stability, substrate binding, protein folding and proton
relay mechanisms. In the catalytic cycles of metalloenzymes, often intramolecular hydrogen
bonding and protonation are important for the efficient generation of high-valent metal-oxo
intermediates ultimately responsible for substrate oxidation (Sono, M.; Roach, M. P. et al
1996, Groves, J. T. 2003, Ortiz de Montellano, P. R. 2004, Kadish, K. M.; Smith, K. M. et al
2010, de Visser, S. P.; Kumar, D. 2011, Dunford, H. B. 1999, Poulos, T. L. 2000, Veitch, N.
C. Smith, A. T. 2000, Kovaleva, E. G.; Neibergaul, M. B. et al 2007, Korendovych, I. V.;
Kryatov, S. V. et al 2007). For example, proton delivery to the coordinated peroxo moiety in
metal monooxygenase hydroxylase, a dinuclear non-heme iron enzyme, triggers O–O bond
cleavage in peroxo intermediate P and affords formation of a high-valent diiron-oxo
intermediate Q, which is capable of hydroxylating methane, the native substrate of MMO
(Murray, L. J.; Lippard, S. J. 2007, Siegbahn, P. E. M.; Blomberg, M. R. A. 2010). Similar
effects operate in heme systems, where the exact location of a proton-donating functionality
in myoglobin mutants can convert an oxygen carrier into a peroxidase (Raven, E. L. 2003,
Watanabe, Y.; Nakajima, H. et al 2007). Perhaps the best studied example of non-covalent
secondary sphere interactions and their role in redox chemistry of the enzyme is provided by
the cytochromes P450.
The cytochromes P450 (P450s) are vital enzymes for human health that are found in a vast
number of biosystems ranging from eukaryotes to prokaryotes (Sono, M.; Roach, M. P. et al
222
1996, Groves, J. T. 2003, Ortiz de Montellano, P. R. 2004, Kadish, K. M.; Smith, K. M. et al
2010). They catalyze a large selection of oxygen atom transfer reactions including substrate
hydroxylation (aliphatic as well as aromatic), epoxidation, sulfoxidation and dehydrogenation
(Groves, J. T.; Shalyaev, K. et al 2000). As a consequence, these enzymes are dominant in
the liver, where they participate in the metabolism of drugs and xenobiotics as well as
detoxification processes (Guengerich, F. P. 2001, Munro, A. W.; Girvan, H. M. et al 2007).
In addition, they have functions in the area of biosynthesis of, e.g., hormones. The P450s
contain a heme-based active site where the metal (iron) is attached to the protein via a thiolate
linkage with a cysteinate residue. The sixth ligand site of the metal is the oxygen coordination
site and using a catalytic cycle that includes two reduction and two protonation steps and the
binding of molecular oxygen on a heme center an iron(IV)-oxo heme cation radical species is
created that is also called Compound I (CpdI) (Sono, M.; Roach, M. P. et al 1996, Groves, J.
T. 2003, Ortiz de Montellano, P. R. 2004, Kadish, K. M.; Smith, K. M. et al 2010, de Visser,
S. P.; Kumar, D. 2011, Denisov, I. G.; Makris, T. M. et al 2005, Nam, W. 2007). Due to its
high reactivity and consequently short lifetime, experimental characterization of CpdI was
hampered for a long time and only recently Rittle and Green found ways to spectroscopically
detect it and study its reactivity patterns (Rittle, J.; Green, M. T. 2010). Figure 8.1 shows the
substrate bound resting state structure of P450cam as taken from the 1DZ9 protein databank
(pdb) file (Schlichting, I.; Berendzen, J. et al 2000). The oxygen atom is in close proximity
to the substrate but also interacts with several nearby hydrogen bond donors, such as the
hydroxyl group of Thr252. Although it is believed that hydrogen bonding donors, such as the
alcohol group of Thr252 are vital for efficient proton transfer during the catalytic cycle, it is
not known if this hydrogen bond affects the subsequent substrate hydroxylation.
223
Figure 8.1: Active site structure of P450 with key amino acids and substrate (camphor) and
solvent water (W) highlighted. Amino acids labelled as in the pdb file.
To gain further insight into the chemical and physical properties of iron(IV)-oxo species,
biomimetic model complexes have been designed that mimic the activity and structure of key
enzyme intermediates (Costas, M.; Mehn, M. P. et al 2004, Kryatov, S. V.; Rybak-Akimova,
E. V. et al 2005, Abu-Omar, M. M.; Loaiza, A. et al 2005, van Eldik, R. 2007, Bruijnincx, P.
C. A.; van Koten, G. et al 2008). These studies have given insight into the effect of the heme,
the nature of the axial ligand and the substrate range of CpdI and related high-valent iron-oxo
species. Very little is known, however, regarding the effects of hydrogen bonding interactions
on the distal site of CpdI and the role of donating hydrogen bonds on the reactivity of the
metal-oxo groups in general. It may be envisaged that these hydrogen bonding interactions
affect the chemical and electronic properties of the oxidant and its reactivity patterns. In line
with this, environmental effects were shown to be important in biomimetic metal-oxo species.
Cys357
Thr252
W902
camphor
224
Recent studies showed dramatic effects on the electronic properties and catalytic activity of
metal-oxo oxidants upon addition of redox inactive cations, e.g. Sc3+
, Ca2+
or Zn2+
, to a
reaction mixture (Pfaff, F. F.; Kundu, S. et al 2011, Fukuzumi, S.; Morimoto, Y. et al 2010).
In particular, addition of Zn2+
to a manganese(V)-oxo corrolazine system changed the
electronic state to a manganese(IV)-oxo corrolazine cation radical and improved its catalytic
properties through valence tautomerism (Leeladee, P.; Baglia, R. A. et al 2012). Although,
biomimetic studies on distal hydrogen bonding centers are scarce there have been some
recent reports on metal-oxo models with hydrogen bond donating ligands (Tani, F.; Matsu-
ura, M. et al 2001, Bénisvy, L.; Halut, S. et al 2006, Brook, R. L.; Borovik, A. S. 2010). In
one of these studies, Borovik and co-workers focussed on metal-oxo complexes,
[MIV
(O)(H3buea)]– with M = Fe, Mn and H3buea = tris[N′-tert-butylureayl-N-ethylene]amine,
as displayed in Scheme 8.1 (MacBeth, C. E.; Gupta, R. et al 2004, Borovik, A. S. 2005,
Parsell, T. H.; Behan, R. K. et al 2006, Shook, R. L.; Borovik, A. S. 2010, Gupta, R.; Lacey,
D. C. et al 2012). Thus, the distal hydrogen bonding interactions in [MnIV
(O)(H3buea)]– were
shown to affect the basicity of the metal-oxo complex considerably and as a result the
manganese(IV)-oxo complex reacts with substrates with stronger C–H bond strengths via
hydrogen atom abstraction than the corresponding manganese(III)-oxo system (Parsell, T. H.;
Yang, M.-Y. et al 2009). Furthermore, intramolecular hydrogen bonding toward the oxo
group stabilized the iron(IV)-oxo species by about 6 kcal mol–1
but more importantly,
decreased the oxygen atom transfer ability dramatically, and made it a somewhat weaker
oxidant (Dey, A.; Hocking, R. K. et al 2006). Further studies showed these complexes to be
highly reactive and even catalytically converted O2 to water in a mechanism resembling that
of the enzyme cytochrome c oxidase (Shook, R. L.; Peterson, S. M. et al 2011). Recent
studies of Kass and co-workers (Shokri, A.; Abedin, A. et al 2012) showed that
intramolecular hydrogen bonding can dramatically influence the pKa values of molecules, and
225
thereby play a key role in reactivities. Clearly, intramolecular hydrogen bonding interactions
to a metal-oxo group affect the catalytic properties of the oxidant in either positive or
negative way. However, the intricate details of this process are currently unknown, and,
therefore, warrant a computational study that gains insight into the effect of distal hydrogen
bonding processes in nonheme iron(IV)-oxo and manganese(IV)-oxo oxidants. As such, we
decided to do a density functional theory (DFT) study into the electronic properties of
[MIV
(O)(H3buea)]– with M = Fe, Mn and H3buea = tris[N′-tert-butylureayl-N-ethylene]amine,
Scheme 8.1, RFe/RMn. This is a tripodal bipyramidal compound with three donating hydrogen
bonds to the oxo group. To find the effect of the substituent on the kinetics of the H-atom
abstraction reaction we also did studies where the three tert-butyl groups were replaced by
either smaller alkyl chains or a hydrogen atom, which are identified as reactants RX with X =
H, Me, Et or i-Pr. The studies presented here highlight the effect of hydrogen bonding
interactions on the distal site of transition metal-oxo complexes and the influence this has on
the kinetics and thermodynamics of hydrogen atom abstraction reactions.
Scheme 8.1: Models studied in this work.
N
N MIV
N
NO
N
N
NX
X
H
H
HX
O
O
O
M = Mn, FeX = H, Me, Et, i-Pr, t-Bu RH: M = Mn, X = H
RMe: M = Mn, X = MeREt: M = Mn, X = EtRi-Pr: M = Mn, X = i-Pr
RMn: M = Mn, X = t-BuRFe: M = Fe, X = t-Bu
226
8.2 Methods
The studies presented in this work use density functional theory (DFT) methods as
implemented in the Gaussian-03 and Gaussian-09 software packages (Frisch, M. J. et al
2003). We use the unrestricted B3LYP density functional method (Becke, A. D. 1993, Lee,
C.; Yang, W. et al 1988) combined to a basis set (B1) that is build up from LACVP on Mn
and Fe and 6–31G on the rest of the atoms (Karamzadeh, B.; Kumar, D. et al 2010). All
structures described in this work are the result of a full geometry minimization (without
constraints) using these methods and a subsequent frequency calculation characterized the
structures as local minima or first order saddle points (transition states) with either real
frequencies only or a single imaginary frequency for the correct mode. Frequencies reported
in this work were scaled with a value of 0.9257 (Hay, P. J.; Wadt, W. R. 1985). Transition
states were located by initially running a stepwise geometry scan between two local minima
by changing the reaction coordinate in a full geometry optimization with one degree of
freedom fixed. The maximum point of these geometry scans was used as a starting point for
the transition state optimizations and established that the transition state is indeed connected
to the minima on each side of the barrier. These methods have been shown to accurately
reproduce structures and free energies of activation of reaction processes of metal-oxo
reactivities (Kumar, D.; Thiel, W. et al 2011). Further improvement of the energetics was
obtained through single point calculations using an LACV3P+ basis set on Mn and Fe and 6–
311+G* on the rest of the atoms: basis st B2. All energies reported in this work were taken
from the UB3LYP/B2 calculations and were corrected for ZPE. Single point calculations
with dispersion corrected B3LYP (Schwabe, T.; Grimme, S. 2007) were performed in
Gaussian and confirmed the obtained trends. Free energies reported here use UB3LYP-D/B2
dispersion corrected energies and include entropic, thermal and solvent corrections to the
energy at 298 K. To test the effect of the environment on the barrier heights we did single
227
point calculations on all structures using the polarized continuum model as implemented in
Gaussian mimicking an acetonitrile solution, however, these studies only gave minor changes
in relative energies as compared to the gas-phase results and no changes in the spin state
ordering. Free energies contain entropic and thermal corrections to the energy at 298 K and
also use energies calculated with basis set B2. The described methods were applied
previously to oxygen atom transfer processes of metal-oxo complexes and calculated free
energies of activation were found to be within 3 kcal mol–1
of experimental data (Kumar, D.;
de Visser, S. P. et al 2005, de Visser, S. P.; Oh, K. et al 2007, Vardhaman, A. K.; Sastri, C.
V. et al 2011).
8.3 Results
In order to understand the effect of distal hydrogen bonding to metal(IV)-oxo intermediates,
we investigated the intrinsic chemical properties as well as the reactivity patterns of
[MIV
(O)(H3buea)]– with M = Mn or Fe, i.e. RMn and RFe. Before we discuss the reactivity
patterns, we will start, however, with an in-depth study into the low-lying electronic states of
these complexes. Figure 8.2 shows the high-lying occupied and virtual orbitals of RMn as
taken from the optimized quartet spin geometry. The molecular orbitals shown in Figure 8.2
are dominated by the metal 3d atomic orbitals and their interactions with neighboring atomic
orbitals. The molecular valence orbitals include a set of * orbitals (*xy, *xz, *yz) and a
pair of * orbitals (*x2–y2, *z2), where the z-axis is aligned with the metal-oxo bond.
228
Figure 8.2: High-lying occupied and low-lying virtual orbitals of 4RMn. Orbital energies are
reported in au.
The *xz and *yz orbitals represent antibonding interactions between the metal 3dxz/3dyz and
2px/2py orbitals on oxygen. The *xy orbital is built up from antibonding interactions of the
metal 3dxy orbital with 2p orbitals of ligands in the xy-plane of symmetry. In heme enzymes
this orbital is non-bonding, but in pentacoordinated complexes like structures R here, there
are interactions with nitrogen atoms in the xy-plane (Hirao, H.; Que Jr, L. et al 2008). As a
consequence, it is substantially higher in energy than the corresponding orbital in heme based
iron(IV)-oxo species. Indeed, in 4RMn it is even higher in energy than the *xz and *yz
orbitals by more than 10 kcal mol–1
, which is unusual as compared to other heme and
nonheme metal(IV)-oxo calculations, where the *xy orbital tends to be the lowest in energy
(Kumar, D.; Hirao, H. et al 2005, de Visser, S. P. 2006, Ye, S.; Neese, F. 2011, Shaik, S.;
Kumar, D. et al 2005). Thus, in octahedral symmetry the 3dxy atomic orbital cannot interact
with orbitals in the xy-plane and hence is non-bonding (xy). By contrast, with only three
ligands in the xy-plane in trigonal bipyramidal structures there are interactions between the
3dxy atomic orbital on the metal with 2p orbitals on the ligands and hence the molecular *xy
2*z
xz* yz*
xy*
–0.083
–0.108–0.101
0.011
0.069
xy*
xz*yz*
22*yx
22*yx
2*z
0.073
0.0570.062
0.080
0.127
xy*
xz*yz*
22*yx
2*z
-orbitals -orbitals
229
orbital is raised in energy with respect to heme enzymes. The * orbitals of
[MIV
(O)(H3buea)]– are high in energy and virtual, and look similar in shape and appearance
as those calculated before for nonheme metal-oxo complexes (Kumar, D.; Hirao, H. et al
2005, de Visser, S. P. 2006, Ye, S.; Neese, F. 2011, Shaik, S.; Kumar, D. et al 2005).
Interestingly, the orbitals displayed in Figure 8.2 give very little electron density
contributions on the bridging N–H bonds, which seems to implicate that their electronic
effect on the molecular orbitals is small. The electronic ground state of 4RMn has orbital
occupation *xz1 *yz
1 *xy
1 and formally corresponds to a manganese(IV)-oxo complex.
This state is well separated from the lowest lying doublet and sextet spin states 2RMn and
6RMn by 17.6 and 14.2 kcal mol
–1, respectively (E+ZPE in solvent; basis set BS2). The spin
state ordering and relative energies change little when the effect of either solvent, counterions
or a change in DFT method is applied. These spin state orderings are in agreement with
experimental EPR studies that characterized [MnIV
(O)(H3buea)]– as a high spin ground state
(Shook, R. L.; Borovik, A. S. 2010).
Replacing the manganese with iron gives only minor changes to the molecular orbitals and
electronic configuration and retains a high-spin ground state with orbital occupation *xz1
*yz1 *xy
1 *x2–y2
1. Experimental electron paramagnetic resonance studies indeed identified
this complex as a high-spin species in agreement with what we find here (Shook, R. L.;
Borovik, A. S. 2010). This result contrasts biomimetic iron(IV)-oxo complexes that are
usually have a triplet spin ground state (Rohde, J.-U.; In, J.-H. et al 2003, Martinho, M.;
Banse, F. et al 2005, Sastri, C. V.; Seo, M. S. et al 2005, De Oliveira, F. T.; Chanda, A. et al
2007, Jackson, T. A.; Rohde, J.-U. et al 2008). On the other hand, most enzymatic nonheme
iron(IV)-oxo species have a high-spin (quintet) ground state instead (Price, J. C.; Barr, E. W.
et al 2003, Hoffart, L. M.; Barr, E. W. et al 2006, Galonić, D. P.; Barr, E. W. et al 2007). This
apparent disparity in triplet-quintet spin state ordering and relative energies of nonheme
230
iron(IV)-oxo complexes was proposed to be correlated with the *xy/*x2–y2 orbital energy
difference and a stabilization of the *xy orbital led to a triplet spin ground state (Hirao, H.;
Que Jr, L. et al 2008). The data shown in Figure 8.2 show that the *xy orbital, in contrast to
other nonheme iron(IV)-oxo complexes in the literature, is higher in energy than the *xz and
*yz molecular orbitals. Since, the *xy and *x2–y2 orbitals are in the same plane of
symmetry, their energy separation affects the spin state ordering. In pentacoordinated metal
complexes both orbitals have similar amount of antibonding interactions and consequently
they are of similar energy, thereby stabilizing the quintet spin state dramatically. Thus, in
trigonal bipyramidal conformation the 3dxy atomic orbital on the metal in the metal(IV)-oxo
complex will be able to form bonding and antibonding interactions with atoms in the xy-
plane. These interactions are not dramatically different for the 3dx2–y2 orbital and hence the
*xy and *x2–y2 orbitals are close in energy for structures R. Metals with four ligands in the
xy-plane, by contrast, have 3dxy and 3dx2–y2 orbitals that are either aligned with the metal-
ligand bonds or are in between the metal-ligand bonds. Hence, the energy gap between *xy
and *x2–y2 is wider in hexacoordinated metal complexes as compared to pentacoordinated
complexes. As a result of that spectroscopic studies on a trigonal bipyramidal iron(IV)-oxo
species give these species as either intermediate (triplet) or high spin state (England, J.; Guo,
Y. et al 2011, Wong, S. D.; Bell III, C. B. et al 2011), dependent on the ligand system.
Optimized geometries of 4RMn and
5RFe are given in Figure 8.3 and are in good agreement
with previous calculations on related complexes (Gupta, R.; Lacey, D. C. et al 2012). The
metal-oxo bonds are long, i.e. 1.706 Å for the Mn–O bond in 4RMn and 1.683 Å for the Fe–O
bond in 5RFe. This is typical for high-spin metal-oxo complexes and, for instance, previous
studies on Mn(IV)-oxo porphyrins also found distances around 1.7 Å for quintet spin state
structures (Balcells, D.; Raynaud, C. et al 2008, Latifi, R.; Tahsini, L. et al 2011, de Visser,
S. P.; Ogliaro, F. et al 2001). The calculated Mn–O distance is dependent on the overall spin
231
state and oxidation state of the metal, and for instance Mn(V)-oxo complexes give
considerably shorter Mn–O distances of around 1.54 Å (Prokop, K. A.; Neu, H. M. et al
2011). Our optimized geometry of 5RFe is similar to related models calculated before by other
groups and also matches typical high-spin iron(IV)-oxo distances that also found bond
lengths around 1.68 Å (Dey, A.; Hocking, R. K. et al 2006, Rohde, J.-U.; In, J.-H. et al 2003,
Martinho, M.; Banse, F. et al 2005, Sastri, C. V.; Seo, M. S. et al 2005, De Oliveira, F. T.;
Chanda, A. et al 2007, Jackson, T. A.; Rohde, J.-U. et al 2008, de Visser, S. P.; Shaik, S. et al
2003, Siegbahn, P. E. M.; Borowski, T. 2006). This implies that donating hydrogen bonding
interactions toward the metal(IV)-oxo group, have little effect on the metal(IV)-oxo bond
length and bond strength. Green showed that the metal-oxo distance correlates with the
metal-oxo vibration according to Badger’s rule (Green, M. T. 2006), hence little
spectroscopic differences are expected due to internal hydrogen bonding interactions. Our
calculations give a Mn–O stretch vibration of 714 cm–1
, which compares favourably with the
reported FT-IR value from Ref (Parsell, T. H.; Behan, R. K. et al 2006). Furthermore,
replacing MnIV
-16
O by MnIV
-18
O gives a decrease of the manganese-oxo stretch vibration by
17 cm–1
as compared to the experimentally reported value of –18 cm–1
. Therefore, the
calculations correctly reproduce experimental spectroscopic features.
Figure 8.3: Optimized geometries with bond lengths in angstroms of 4RMn and
5RFe; group
spin densities () taken from UB3LYP/BS2 calculations. Also given are relative energies
rM(IV)O = 1.706 / 1.683
rM(IV)Nax = 2.207 / 2.152
M = 2.86 / 3.25
O = 0.24 / 0.43
4RMn / 5RFe
4RMn (2RMn) [6RMn]: 0.0 (22.1) [14.2]5RFe (3RFe) [1RFe]: 0.0 (17.0) [26.6]
E + ZPE + Esolv
232
(including ZPE and solvent corrections in kcal mol–1
) of all low lying spin states for RMn and
RFe.
Subsequently, we investigated the hydrogen atom abstraction mechanism of substrates by
2,4,6RMn and
1,3,5RFe using dehydroanthracene (DHA) and 1,4-cyclohexadiene (CHD) as
model substrates. The reactions are stepwise with two subsequent hydrogen atom abstraction
processes leading to dehydrogenated products, i.e. anthracene and benzene, respectively. The
initial hydrogen atom abstraction is rate determining and passes a transition state (TSH) to
form a metal(III)-hydroxo complex (I) and is followed by a dehydrogenation transition state
(TSdh) leading to water and dehydrogenated product (P). Figure 8.4 displays the mechanism
for the initial hydrogen atom abstraction step by RMn using DHA as a substrate (left-hand-
side), while the optimized geometries of the H-abstraction transition states for both DHA
(TSH,DHA) and CHD (TSH,CHD) are given on the right-hand-side of Figure 8.4. The rest of the
mechanisms found for CHD is similar and only the energies vary slightly. The reaction takes
place via single-state-reactivity on dominant quartet (Mn) and quintet (Fe) spin state surfaces.
During the complete mechanism these spin states remain the ground state and other spin
states are well higher in energy, therefore, we will focus on the lowest lying spin state results
only. We also calculated the reaction mechanisms starting from 2RMn,
6RMn,
1RFe and
3RFe.
However, the barrier heights and the local minima for these alternative spin state surfaces
were considerably higher in energy than those reported for the structures found in the quartet
spin state for the manganese complex and the quintet spin state for the iron complex. As a
consequence, only the quartet spin state of RMn and the quintet spin state of RFe are
catalytically relevant, i.e. the reaction proceeds via single-state reactivity. This is unusual for
metal-oxo reactivities since often multiple spin states are involved in reaction mechanisms
and the intricate surface crossings between these spin states determines the actual mechanism
(Shaik, S.; de Visser, S. P. et al 2002, Manner, V. W.; Lindsay, A. D. et al 2012). However,
233
extensive previous studies on nonheme iron(IV)-oxo oxidants showed that generally the
barriers on the quintet spin state surface are the lowest in energy (Kumar, D.; Hirao, H. et al
2005, de Visser, S. P. 2006, Ye, S.; Neese, F. 2011, Shaik, S.; Kumar, D. et al 2005, Aluri,
S.; de Visser, S. P. 2007, de Visser, S. P.; Tahsini, L. et al 2009, de Visser, S. P.; Latifi, R. et
al 2011). Thus, on the quintet spin state an electron is transferred from the substrate into the
*z2 molecular orbital, which has been termed a 5 process. By contrast, the lowest lying
triplet spin mechanism contains an electron transfer into the *xz orbital, labelled as a 3
process (Hirao, H.; Kumar, D. et al 2006, de Visser, S. P. 2006). Generally, mechanisms via a
5 pathway are lower in energy than those passing a 3 state. Since the structures discussed in
this work already start off from a quintet spin state, no further spin state crossing is needed
and the reaction proceeds on the same spin state surface via the favourable 5 mechanism.
The first H-atom abstraction from substrate by the oxidant leads to either a Mn(III)-hydroxo
or Fe(III)-hydroxo complex, whereby the electron transfer from the substrate enters the metal
*z2 orbital to give an electronic state for 4IMn of *xz
↑ *yz
↑ *xy
↑ *z2
↑ Sub
↓ and for
5IFe of
*xz↑ *yz
↑ *xy
↑ *z2
↑ *x2–y2
↑ Sub
↓, whereby the Sub orbital represents the radical on the
substrate rest group. Thus, the metal 3d-block is largely exchange coupled with either four
(Mn) or five (Fe) unpaired electrons, whereas the radical on the substrate moiety (Sub)
contains a down-spin electron. We also found an alternative transition state (5TSH,Fe’) with
configuration *xy2 *xz
↑ *yz
↑ *z2
↑ Sub
↑, however, in the gas-phase this barrier was 5.2 kcal
mol–1
higher in energy than the one reported in Figure 8.4. Therefore, electron transfer from
the substrate into the virtual *x2–y2 orbital is energetically favourable over electron transfer
in the low-lying *xy molecular orbital as it gives a fully exchange coupled metal 3d-block.
As follows from Figure 8.4, the H-atom abstraction from DHA by 4RMn costs 14.8 kcal mol
–1
in the gas-phase, whereas for the mechanism starting from 5RFe it is 16.6 kcal mol
–1. The H-
abstraction from CHD has a much higher barrier for 5RFe than for
4RMn, but the effect of
234
solvent reverses the trends and makes the iron complex a slightly better oxidant by 2.2 kcal
mol–1
. These relative barrier heights implicate that RFe is a slightly better oxidant than RMn.
The results contrast the reactivity patterns of manganese-porphyrin versus iron-porphyrin,
where generally more favourable reactions are observed with manganese based oxidants
(Groves, J. T.; Stern, M. K. 1988, Yoshizawa, K.; Shiota, Y. et al 1998, Kurahashi, T.;
Kikuchi, A. et al 2008). The reason for this difference is that manganese-porphyrins,
manganese-corroles and manganese-corrolazines tend to have a low-spin ground state and the
corresponding Mn(V)-oxo species are closed-shell singlets (Nehru, K.; Kim, S. J. et al 2007,
Song, W. J.; Seo, M. S. et al 2007, Prokop, K. A.; de Visser, S. P. et al 2010, Kumar, A.;
Goldberg, I. et al 2010). By contrast, iron(IV)-oxo porphyrin cation radical species have
degenerate doublet and quartet spin states. As has been shown in many computational
studies, high-spin states react with substrates with lower H-abstraction barriers than
intermediate or low-spin states (Shaik, S.; Chen, H. et al 2011).
Figure 8.4: (a) Potential energy profile of hydrogen atom abstraction from DHA by 4,6,2
RMn
as calculated using DFT methods with energies in kcal mol–1
relative to the quartet spin
reactant complex. Energies are taken from the UB3LYP/BS2 calculations in the gas-phase,
while solvent corrected values are in parenthesis. Free energies with solvent, entropic and
dispersion corrections are given in square brackets. (b) Optimized geometries of the transition
2IMn
4TSH,Mn (5TSH,Fe)
4RMn
4IMn
4TSH,Mn
0.0 (0.0) [0.0]
14.8 (15.2)
[3.2]
rMO = 1.806 (1.815)
rHO = 2.948
(3.385)
rOH = 1.249 (1.322)
rHC = 1.346 (1.270)
i1822.6 (i1595.7) cm–1
4TSH,CHD,Mn [5TSH,CHD,Fe]
rMO = 1.803 [1.803]
rOH = 1.316 [1.179]
rHC = 1.295 [1.425]
i1667.4 [i1224.3] cm–1
E+ZPE = 15.8 (14.9) [22.7 (12.7)]
24.7 (23.6)
34.6 (34.6) [21.8]
6RMn
2RMn
15.4 (15.1) [13.5]
16.5 (17.6) [19.0]
6TSH,Mn
2TSH,Mn
–10.1 (–6.4) [–13.9]
–10.2 (–6.4) [–14.3]
10.0 (12.4) [1.7]
6IMn
(a) (b)
235
states for hydrogen atom abstraction from DHA and CHD with bond lengths in angstroms
and the imaginary frequency in the transition state in cm–1
.
Optimized geometries are typical for nonheme metal-oxo complexes and are similar to
previous studies in the field (Hirao, H.; Que Jr, L. et al 2008, Kumar, D.; Hirao, H. et al 2005,
de Visser, S. P. 2006, Ye, S.; Neese, F. 2011, Shaik, S.; Kumar, D. et al 2005, Hirao, H.;
Kumar, D. et al 2006, de Visser, S. P. 2006, Johansson, A. J.; Blomberg, M. R. A. et al 2007,
Bernasconi, L.; Baerends, E. J. 2008, Comba, P.; Maurer, M. et al 2009, Prat, I.; Mathieson,
J. S. et al 2011). With DHA as a substrate the transition states are product-like for
4TSH,Mn,DHA with longer C–H than O–H bonds, but early for
5TSH,Fe,DHA: rCH = 1.346 (1.270)
and rOH = 1.249 (1.322) Å for 4TSH,Mn,DHA, respectively. Much larger differences are found
between the two H-atom abstraction transition states with CHD as a substrate, whereby the
Mn-structure has a central H-atom (rCH = 1.295 and rOH = 1.316 Å), but a more product-like
geometry is found for 5TSH,Fe,CHD (rCH = 1.425 and rOH = 1.179 Å). All these transition states
are characterized with large imaginary frequencies from i1224.3 – i1822.6 cm–1
, which
indicate that the reactions will encounter a large kinetic isotope effect for the replacement of
hydrogen atoms by deuterium as found before on related metal-oxo reactivities (Kamachi, T.;
Yoshizawa, K. 2003, Kumar, D.; de Visser, S. P. et al 2003, Kumar, D.; de Visser, S. P. et al
2004, D. Kumar, de Visser, S. P. et al 2004).
To test the effect of the substituents on the donating distal hydrogen bond, we did a further
set of calculations on the [MnIV
(O)(H3buea)]– complex where the t-butyl groups of H3buea
were replaced by i-propyl (Ri-Pr), ethyl (REt), methyl (RMe) or hydrogen (RH). We initially did
a full geometry optimization of the reactant complexes of RX, X = i-Pr, Et, Me and H in the
quartet spin state and the optimized geometries are given in Figure 8.5 alongside those for
4RMn that contains X = t-Bu. Geometrically, no dramatic changes are observed when the
substituent group is gradually reduced from t-butyl to a hydrogen atom. The Mn–O distance
236
varies by a maximum of 0.012 Å, whereas the bond length of the metal with the axial ligand
Mn–Nax varies by 0.026 Å. Similar variations are found for the donating hydrogen bond
distances. A look at the group spin densities of these complexes shows minor differences in
electronic configuration and unpaired spin density.
Figure 8.5 also gives the vibrational frequencies of the Mn–O distances in all complexes,
MnO. As can be seen there are strong variations in the Mn–O stretch vibration from a value of
701.7 cm–1
for 4Ri-Pr to 731.4 cm
–1 for
4RH.
Figure 8.5: Optimized geometries of 4RX structures with different substituents X with bond
lengths in angstroms and the vibrational frequency in wave numbers. Note that 4RMn has X =
t-Bu.
OH
H
H
N
N
N
rMnO = 1.706 / 1.718 / 1.717 / 1.717 / 1.709rMnNax = 2.207 / 2.205 / 2.201 / 2.202 / 2.181
rOH = 1.780 / 1.825 / 1.820 / 1.822 / 1.804
rOH = 1.718 / 1.731 / 1.745 / 1.739 / 1.738rOH = 1.769
/ 1.780/ 1.780/ 1.790/ 1.788
X
X
X
4Rt-Bu / 4Ri-Pr / 4REt / 4RMe / 4RH
MnO = 714.4 / 701.7 / 706.4 / 712.2 / 731.4 cm–1
237
Figure 8.6: (a) Barrier heights (E‡+ZPE in solvent) for H-atom abstraction from DHA by
various complexes manganese based complexes RX (X = t-Bu, i-Pr, Et, Me or H). (b) Barrier
heights as a function of the Mn–O frequency (MnO) in the reactant complex.
Subsequently, we investigated the H-atom abstraction of DHA by each of the oxidants
displayed in Figure 8.5. Similar to what we reported above, the reactions are stepwise with an
initial hydrogen atom abstraction to form a radical intermediate. DFT calculated barrier
heights (with solvent corrections included) on the quartet spin state are given in Figure 8.6 for
comparison. The lowest H-abstraction barrier is found with i-propyl substituents (E‡+ZPE =
13.4 kcal mol–1
). This barrier gradually increases to 16.1 kcal mol–1
upon replacement of the
i-propyl groups with shorter alkyl chains or a hydrogen atom. Consequently, there is a small
but significant substituent effect on the H-atom abstraction processes by these metal(IV)-oxo
complexes. Enlarging the substituent group further to t-butyl reverses the trend and raises the
barrier heights again. This is probably due to stereochemical interactions of the atoms of the
substituents.
In order to understand the trend in barrier heights, we plot in the lower panel of Figure 8.6
the barrier height of the hydrogen atom abstraction reaction relative to the MnO frequency in
the reactants. An almost linear correlation is found, which implicates that the strength of the
Mn–O bond affects the hydrogen atom abstraction. Interestingly, the complex with the longer
y = 0.09x - 48.47
R² = 0.91
0
5
10
15
20
25
690 700 710 720 730 740 750
E‡+ZPE
[kcal mol–1]
MnO [cm–1]
0
5
10
15
20
E‡+ZPE
[kcal mol–1]15.2
13.414.1
15.016.1
t-Bu i-Pr Et Me H
238
Mn–O bond length in the reactants, and consequently the system with the weakest Mn–O
bond strength, gives the highest reaction barrier for this series of hydrogen atom abstractions.
Thus, even though the Mn–O bond is broken in the substrate hydroxylation/dehydrogenation
reaction, it is not its binding energy that determines the rate constant of hydrogen atom
abstraction. Instead, as will be explained in the Discussion section, it is the pKa value of the
oxo group and the electron affinity of the oxidant.
8.4 Discussion
This work describes the electronic and catalytic properties of metal(IV)-oxo complexes
with distal hydrogen bonding interactions. In the following we will try to rationalize the
effect of distal hydrogen bonding interactions on the mechanism and kinetics of H-atom
abstraction reactions by metal(IV)-oxo oxidants. Experimental studies (Mayer, J. M. 1998,
Mayer, J. M. 2004, Mader, E. A.; Manner, V. W. et al 2009, Kaizer, J.; Klinker, E. J. et al
2004, Yoon, J.; Wilson, S. A. et al 2009, Lansky, D. E.; Goldberg, D. P. 2006, Bell, S. R.;
Groves, J. T. 2009) on many occasions have shown that the rate constant of a H-atom
abstraction reaction correlates linearly with the strength of the C–H bond of the substrate that
is broken. Generally, a plot of the natural logarithm of the rate constant of hydrogen atom
abstraction of a selection of substrates by a specific oxidant gives a linear correlation with the
strength of these C–H bonds, i.e. bond dissociation energy (BDECH). Density functional
theory studies in combination with valence bond descriptions showed this correlation to
originate from the relative electronic configurations of the reactants and products (Shaik, S.;
Kumar, D. et al 2008, Shaik, S.; Lai, W. et al 2010, Latifi, R.; Bagherzadeh, M. et al 2009,
Kumar, D.; Karamzadeh, B. et al 2010, Kumar, D.; Sastry, G. N. et al 2011, Kumar, D.;
Sastry, G. N. et al 2012, de Visser, S. P.; Kumar, D. et al 2004). Further studies highlighted
the importance of the strength of the O–H bond that is formed in the processes and essentially
239
assigned this O–H bond dissociation energy (BDEOH) of the metal(III)-hydroxo product as a
mimic to assign the catalytic power of the oxidant (de Visser, S. P. 2010). Recent
electrochemical studies of Fujii and co-workers (Takahashi, A.; Kurahashi, T. et al 2011)
established redox potentials of iron(IV)-oxo porphyrin cation radical models with varying
axial ligands. They found that an anionic axial ligand gave different redox potentials than a
neutral axial ligand, but changing an anionic axial ligand by another anionic axial ligand gave
little changes in redox potential. Subsequently, they investigated reactivity patterns and did
not find correlations between the rate constants of olefin epoxidation with electron properties
of the oxidant including the redox potential, however, they did find correlations with the
redox potential of the iron(III)porphyrin product complex (Takahashi, A.; Yamaki, D. et al
2012). By contrast, Nam and co-worker (Sastri, C. V.; Lee, L. et al 2007) investigated the
axial ligand effect of nonheme iron(IV)-oxo complexes with a TMC or TMCS ligand system
(TMC = 1,4,8,11-tetramethyl-1,4,8,11-tetraazacyclotetradecane; TMCS = 1-mercaptoethyl-
4,8,11-trimethyl-1,4,8,11-tetraazacyclotetradecane) and found the reactivity to be dependent
on the blending of quintet and triplet spin state surfaces. Thus the systems of Nam are in a
triplet spin ground state and during the H-atom transfer a spin state crossing to a more
favourable quintet spin state surface occurs. It was shown that the amount of blending and
spin state transfer affects the rate constants of the reaction. To find out whether the bond
dissociation energy of the O–H bond (BDEOH) in the iron(III)-hydroxo complex is related to
the rate constants and hydrogen atom abstraction barriers, we calculated reaction 1 for the
complexes described in this work.
[MIII
(OH)(L)] → [MIV
(O)(L)] + H• + BDEOH (1)
The calculated BDEOH values for 4RMn (
5RFe) are 81.2 (84.1) kcal mol
–1 in the gas-phase
and 83.9 (83.6) kcal mol–1
in solvent, and if the barrier height of H-atom abstraction is
240
proportional to these values, the rate determining barrier should be very similar for these
oxidants. Indeed, the barrier heights for H-atom abstraction from DHA and CHD in Figure
8.3 above show that they are very similar for the iron and manganese complexes in support of
the relative BDEOH values. Our calculated BDEOH values also compare favourably with
experiment as Borovik and co-workers isolated and kinetically investigated the
[MnIV
(O)(H3buea)]– and [Mn
III(OH)(H3buea)]
– complexes and determined a BDEOH value for
this complex of 89 kcal mol–1
.[42]
The calculated BDEOH values for 4RMn and
5RFe are
somewhat lower than those found for [FeIV
(O)(Por+•
)X] with Por = Porphyrin and X an axial
ligand, such as chloride or thiolate, for which values of around 88 kcal mol–1
were obtained
(de Visser, S. P. 2010). Furthermore, much higher BDEOH values were obtained for nonheme
iron(IV)-oxo complexes of enzymatic as well as biomimetic model complexes with values
well over BDEOH = 90 kcal mol–1
(de Visser, S. P. 2010). These systems react with aliphatic
substrates with substantially lower H-atom abstraction barriers than those reported here for
4RMn and
5RFe in agreement with the relatively low BDEOH values.
To understand the effect of hydrogen bonding on the height of the H-abstraction barriers
we investigated the corresponding BDEOH values of hydrogen bonding and non-hydrogen
bonding metal(IV)-oxo complexes. Thus, we took the optimized geometry of 5RFe and the
corresponding iron(III)-hydroxo complex and rotated the peptide fragments of the H3buea
ligand by about 90 in such a way that no hydrogen bonding interaction to the oxo/hydroxo
group remains and recalculated the BDEOH value. The geometries were then again optimized
but with these fixed dihedral angles and the new non-hydrogen bonding energies were
calculated. This way, without hydrogen bonding interactions the [FeIV
(O)(H3buea)] system
has a BDEOH of 94.4 kcal mol–1
. Consequently, hydrogen bonding toward the oxo group in
iron(IV)-oxo complexes lowers the BDEOH value of the oxidant with 10.1 kcal mol–1
. Thus, if
BDEOH correlates linearly with hydrogen atom abstraction barrier as reasoned before (de
241
Visser, S. P. 2010) this would imply a raise in barrier upon addition of hydrogen bonding
interactions at the distal position of the metal(IV)-oxo group by several kcal mol–1
. Indeed, a
comparison of a hydrogen atom abstraction barrier of RFe with the iron(IV)-oxo species of
taurine/-ketoglutarate dioxygenase gives substantially lower hydrogen abstraction barriers
for the enzyme (Latifi, R.; Bagherzadeh, M. et al 2009) possibly due to the lesser number of
hydrogen bonding interactions. In enzymatic systems, such as the P450s, CpdI is generated
from binding of O2 on a heme followed by two protonation steps and a reduction. The distal
hydrogen bonding interactions, therefore, are probably essential for proton transfer
mechanisms in order to facilitate production of an iron(IV)-oxo heme cation radical active
species. However, as shown here, the hydrogen bonding interactions that facilitate these
proton relay processes in the enzyme, actually lower the BDEOH values of the oxidant and
consequently lower the oxidative power of the enzyme. Nonheme iron enzymes, by contrast,
utilize -ketoglutarate as a co-substrate to generate an iron(IV)-oxo active species rather than
protonation and electron transfer mechanisms. To find evidence for our hypothesis that distal
hydrogen bonding interactions will lower the catalytic activity of metal(IV)-oxo oxidants, we
investigated a selection of crystal structures of typical nonheme iron dioxygenases and
searched for distal hydrogen bonding interactions, see Figure 8.7. We selected taurine/-
ketoglutarate dioxygenase (TauD), AlkB repair enzymes and cysteine dioxygenase (CDO)
and took an available pdb file for each of these from the protein databank (O’Brien, J. R.;
Schuller, D. J. et al 2003, Ye, S.; Wu, X. et al 2007, Yu, B.; Hunt, J. F. 2009). Thus, TauD is
a nonheme iron dioxygenase with the metal bound to a 2-His/1-Asp facial triad of His99,
Asp101 and His255 and with the help of -ketoglutarate (KG) as a co-substrate and O2 it
generates an iron(IV)-oxo active species, which abstracts a hydrogen atom from substrate
(taurine) to give hydroxylated products. In TauD, KG binds as a bidentate ligand and the
last ligand position (vacant in Figure 8.7) is reserved for molecular oxygen. As can be seen
242
from Figure 8.7 this last ligand position has no hydrogen bonding donors nearby. The only
polar residues in the active site are involved in hydrogen bonding interactions with the
substrate, such as Arg270 and Asn95. In the AlkB repair enzyme the situation is very similar
and the metal is coordinated via a 2-His/1-Asp facial motif with His131, Asp133 and His187.
Also in AlkB the KG group binds as a bidentate ligand and the sixth ligand site of the metal
is occupied by a water molecule. No hydrogen bonding donors are located nearby this sixth
ligand site, although there is an arginine residue in the vicinity that forms a salt bridge with
KG. The third structure displayed in Figure 8.7 represents cysteine dioxygenase (CDO),
which is a nonheme iron enzyme involved in the detoxification of cysteine in the body. It
catalyzes the dioxygenation of the thiolate group of cysteine to cysteine sulfinic acid
products. The pdb file has an iron active site with the metal linked through interactions with
three histidine residues (His86, His88, His140) to the protein and substrate cysteinate binds as a
bidentate ligand via the amine and thiolate groups to the metal. The carboxylate group of
cysteine is involved in hydrogen bonding interactions with several residues including His155
and Tyr157. These residues are out of range of a possible iron(IV)-oxo intermediate and,
therefore, also in CDO there are no visible hydrogen bonding donors to the iron(IV)-oxo
active species. Clearly, the absence of hydrogen bonding donors to the active species of
nonheme iron enzymes implicates that hydrogen bonding has a negative effect on catalysis
and will raise barrier heights and, therefore, in metalloenzymes hydrogen bonding donors to a
metal-oxo active species are lacking unless an essential proton transfer step is required in the
catalytic cycle.
243
Figure 8.7: Extracts of the active site environments of nonheme iron dioxygenases
representing from left-to-right: TauD (1OS7 pdb), AlkB (3I2O pdb) and CDO (2IC1 pdb).
Amino acids are labelled as in the pdb file.
Figure 8.8: (a) Orientation of substrate attack on the metal(IV)-oxo group with angles in
degrees and group spin densities in au. (b) Electron transfer processes and LUMO orbital that
is filled with one electron in the H-abstraction process.
His99 His255
KG
taurine Arg270
Asn95
TauD (1OS7)
Asp133
KGHis131
His187
DNA chain Arg210
AlkB (3I2O) CDO (2IC1)
His86
His140
His88 Cys93
Tyr157His155Cys
Asp101
2*z
Fe-O-H: 120.0 [117.8] Fe-O-H: 130.4 [121.8]
4TSMn,DHA [4TSMn,CHD] 5TSFe,DHA [5TSFe,CHD]
*z2
FeIV
O
e
H
C
(a) (b)
Mn = 3.63 [3.72]
OH = –0.07 [–0.16]
L = –0.18 [–0.15]
Sub = –0.38 [–0.40]
Fe = 3.96 [3.74]
OH = 0.03 [0.15]
L = 0.39 [0.27]
Sub = –0.38 [–0.16]
244
Table 8.1. Calculated barrier heights for hydrogen atom abstraction from
dehydroanthracene and 1,4-cyclohexadiene by metal-oxo complexes (Energies are in
kcal mol–1
).
Complex E‡ + ZPE E
‡ + ZPE + Esolv G
‡ + Esolv Ref
Dehydroanthracene Data: 4[Mn
IV(O)(H3buea)] or
4RMn 14.8 15.2 21.1 This work
4RMn,H 15.6 16.1 19.1 This work
4RMn,Me 14.5 15.0 18.5 This work
4RMn,Et 14.4 14.1 17.7 This work
4RMn,i-Pr 14.4 13.4 16.1 This work
5RFe 16.6 16.5 23.6 This work
1[Mn
V(O)(corrolazine)] 18.7 18.7 31.2 [32c]
1[Mn
V(O)(corrolazine)F]
– 11.3 13.7 24.4 [32c]
1[Mn
V(O)(corrolazine)CN]
– 9.0 13.0 21.8 [32c]
2[Mn
V(O)(corrolazine
+•)]
+ 0.7 8.4 19.3 [25]
2[Fe
IV(O)(corrole
+•)] 4.1 14.7 25.7 [45]
Cyclohexadiene Data: 4RMn 15.8 14.9 21.4 This work
5RFe 22.7 12.7 18.9 This work
2[Fe
IV(O)(corrole
+•)] 5.1 11.5 22.1 [45]
2[Fe
IV(O)(Por
+•)Cl] 8.3 11.0 9.7 [46]
4[Fe
IV(O)(Por
+•)Cl] 9.1 11.4 11.0 [46]
5[Fe
IV(O)(TMC)(CF3COO)]
+ 3.8 7.8 17.4 [44]
5[Fe
IV(O)(TMCS)]
+ 8.2 11.3 21.6 [44]
4[Ni
II(O2)(TMC)]
+ 18.4 14.2 22.4 [47]
Another factor that affects the barrier heights here is the orientation of the substrate. In
particular, the relatively high H-abstraction barriers of 4RMn and
5RFe complexes discussed
above is due to the non-ideal geometric conformation of the transition states. Figure 8.8
highlights key structural features and group spin densities of the transition states to explain
this. Thus, the substrate attacks the metal(IV)-oxo group sideways whether we choose DHA
or CHD as a substrate with Fe–O–H angles of around 120. These angles are very small for
reaction mechanisms that take place via a 5 pathway.
245
In principle, the hydrogen atom abstraction is accomplished with a one electron transfer
from substrate to oxidant. Starting from 4RMn with *xz
1 *yz
1 *xy
1 configuration, there are
five possible pathways, namely whereby each of the metal 3d-type orbitals is filled with one
extra electron. In the so-called -pathways the H-atom abstraction leads to double occupation
of either the *xz, *yz or *xy orbital with a second electron, whereas in the alternative -
pathways either the *x2–y2 or *z2 orbital is filled with one electron. These electron transfer
processes have been shown to determine the optimized geometry of the transition states (de
Visser, S. P. 2006). Thus, the lowest lying hydrogen atom abstraction barriers, 4TSMn,DHA,
4TSMn,CHD,
5TSFe,DHA and
5TSFe,CHD, include an electron transfer from the substrate into the
*z2 orbital that becomes singly occupied in the radical intermediate, hence these reactions
follows a 4/
5 pathway. As the electron transfer leads to donation of an electron into *z2
orbital, this geometrically should lead to an upright transition state structure with an almost
linear Fe–O–H angle of close to 180 as obtained for analogous complexes without the
hydrogen bonding features (de Visser, S. P. 2006). On the other hand, the -pathway
mechanisms take place with an electron transfer into a * orbital and consequently the
substrate will attacks the oxidant under an angle of around 120 to get favourable orbital
overlap between the donor and acceptor of the electron. In nonheme iron(IV)-oxo oxidants
the triplet spin barriers are usually of 3-pathway type and give structures with Fe–O–H
angles of around 120. Our triplet spin transition states indeed show electron transfer into the
metal -system and a geometry with Fe–O–H angle around 120. Starting from 5RFe, we also
located the 5 transition state (
5TSH,Fe’) with configuration *xy
2 *xz
↑ *yz
↑ *z2
↑ Sub
↑, and
found it to be substantially higher in energy than the transition state for the 5 pathway. It is
clear, therefore, that the -pathways for electron transfer are less favourable in energy than
the -pathways and the occupation of *z2 with an electron. Even a change of transition state
246
geometry from an “ideal” conformation with almost linear Fe–O–H–C dihedral angle to a
more bend structure does not destabilize the 5-pathway enough to bring it above the
3/5
pathways.
In Table 8.1 we compare the H-atom abstraction barriers of the hydrogen bonded iron and
manganese complexes discussed in this work with DHA and CHD with similar complexes
calculated before using density functional theory (DFT) (Prokop, K. A.; de Visser, S. P. et al
2010, Hirao, H.; Que Jr, L. et al 2008, Latifi, R.; Valentine, J. S. et al 2012, Kumar, D.;
Tahsini, L. et al 2009, Latifi, R.; Tahsini, L. et al 2011). We have grouped the calculated
reactivity trends of metal-oxo and metal-superoxo complexes with DHA and CHD substrates
in Table 8.1 as a comparison with the reactivity studies reported here. We give calculated
gas-phase barrier heights (E‡+ZPE) as well as solvent corrected barrier heights
(E‡+ZPE+Esolv) and free energies of activation in solvent (G
‡+Esolv). Due to the fact that
the overall charge is not equal in all these complexes some very large variations are observed
between the gas-phase and solvent corrected results. In the discussion we, therefore, will
focus on the free energies of activation in solvent.
The H-atom abstraction of DHA by 4RMn has a free energy of activation that is somewhat
lower in energy than that found for analogous manganese(V)-oxo corrolazine complexes
(Prokop, K. A.; de Visser, S. P. et al 2010), and considerably lower in energy than
[FeIV
(O)(corrole+•
)] and Ni(II)-superoxo complexes. The free energies of activation of CHD
H-abstraction by 4RMn and
5RFe show comparable reactivity with CHD with
[FeIV
(O)(TMC)(CF3COO)]. Of course, the systems decribed in Table 8.1 follow different
electron transfer processes, and, hence correlations are difficult to assign here. However, the
work described above clearly highlights that 4RMn and
5RFe are in the high-spin configuration
due to the fact that these systems are trigonal bipyramical. Because of that the relative
247
energies of the *xy and *x2–y2 molecular orbitals are small and the systems are stabilized in
high-spin states.
High-spin states, such as 4RMn and
5RFe react with substrates via electron abstraction into a
vacant *z2 orbital that is aligned along the metal-oxo axis. Substrate approach to the oxidant,
therefore, will encounter lesser stereochemical interactions with the ligand along this 5
pathway than in alternative spin states, where the substrate approach is more sideways (de
Visser, S. P. 2006). A comparison of the barrier height of H-abstraction from CHD by 4RFe,
5[Fe
IV(O)(TMC)(CF3COO)]
+ and
5[Fe
IV(O)(TMCS)]
+ shows that the free energy of activation
of 4RFe lies exactly in between that found for the other two complexes. The calculations here
implicate that 4RFe is a comparable oxidant to nonheme iron(IV)-oxo complexes with a TMC
ligand system.
248
8.5 Conclusion
In this work we present a series of DFT calculations on nonheme metal(IV)-oxo complexes,
[MIV
(O)(H3buea)]– with M = Fe/Mn. We show that hydrogen bonding interactions to the oxo
group effect the catalytic properties of the oxidant dramatically. Firstly, they restrict the
approach of the substrate and as a consequence the hydrogen atom abstraction on the low-
lying 5-pathway takes place at an almost perpendicular angle to the metal-oxo bond rather
than aligned with it as would be expected from electron transfer mechanisms. Secondly,
hydrogen bonding lowers the BDEOH value and consequently the oxidative power of the
oxidant. Indeed, relatively high hydrogen atom abstraction barriers from dehydroanthracene
and cyclohexadiene are found as compared to alternative iron and manganese based oxidants.
The work has relevance to P450 chemistry and shows that distal hydrogen bonding
interactions may restrict the catalytic properties of the active species, but on the other hand
may stabilize the proton transfer mechanisms that generate Compound I.
249
9.1 Concluding remarks
The presented thesis is strickly focused on primarily: the electronic structure calculations
aimed at studying the reactivity and trends associated with Iron porphyrin complexes.
Although, two out of the five result chapters were not entirely porphyrin but the
characteristics studied from them is directly relatable to Metal Porphyrin complexes. A quick
recarpitulation of the results obtained from each chapter is professed as viz;
The first result chapter detailed an earlier experimental result presented by sankar and co-
workers on the effect of counter anions on spin state ordering on diiron porphyrin complexes;
although their spectroscopic studies shows a considerable distressing of the porphyrin rings
depending on the approaching counter anions, our computational studies shows that there are
variety of external perturbation that could affect the spin state ordering of the complex apart
from the approaching counter anions.
The second result chapter presented within this thesis shows that All olefins undergo
oxygen atom transfer, whereas compounds with low ionization energy also give a certain
degree of hydride transfer and charge transfer reactions; the computational finding was
strongly supported by the experimentally determined reaction rate which correlate linearly
with the ionization potential of the substrate and indicated that the electron transfer from
substrate to oxidant is rate determining.
A chapter was also presented on the effect of axial ligands, namely; chloride verses
acetonitrile. It was shown that neutral axial ligand leads to displacement of the metal from the
plane through the porphyrin ring and results in different orbital interactions between metal
and porphyrin ring as compared to systems with an anionic ligand. The effect of equatorial
ligand on Porpyrin versus TPFPP was also investigated; it is shown that the substituents on
250
the porphyrin ring can guide substrate binding through electrostatic interactions with halide
atoms, which lowers the barrier heights.
A chapter indicating the regioselectivity of aliphatic hydroxylation against desaturation was
also presented. The key steps in the mechanisms were identified. The detailed computational
studies suggests that both regioselectivity is initiated by a hydrogen atom abstraction step
tailed by OH isomerisatin to form either the two pro-hydroxylation radical intermediate or the
pro-desaturation intermediate.
The last two result chapters differ from the earlier chapters discussed in the sense that the
compex studied in the earlier chapters are related as they are either reactivity or trends
associated to metal porphyrin complexes.
However in the next chapter our findings recognize the key features of a synthetic
homogeneous catalyst for alkyl chain growth on Ru2Pt-carbene. Our result shows that a low
energy mechanism of alkyl formation through consecutive CH3 and H
transfer to the
carbene with low free energy of activation and exergonic driving force. We show that the
reverse process is thermochemically hindered and would lead to an excited triplet
configuration. Whereas in the last chapter, a thorough density functional theory analysis was
done on a biomimetic nonheme metal(IV)-oxo complexes, and it was found that hydrogen
bonding interactions to the oxo group effect the catalytic properties of the oxidant intensely.
Even though the studied complex is a non heme, the work has significance to P450 chemistry
and demonstrated that distal hydrogen bonding interactions may restrict the catalytic
properties of the active species, but on the other hand may stabilize the proton transfer
mechanisms that generate Compound I.
251
References
Abatjoglou, A. G., Billig, E., Bryant, D.R. Organometallics, 2010, 3, 923
Abdulsattar, M. A. 2012. "SiGe superlattice nanocrystal infrared and Raman spectra: A
density functional theory study". J. Appl. Phys. 111 (4): 044306.
Abu-Omar, M. M., Loaiza, A., Hontzeas, N. Chem. Rev. 2005, 105, 2227
Adamo, C., Barone, V. J. Chem. Phys. 1999, 110, 6158
Ahlrichs, R., Bär, M., Häser, M., Horn, H., Kölmel, C., Chem. Phys. Lett. 1989, 162, 165
Alcaide, B., Almendros, P., Luna, A. Chem. Rev. 2009, 109, 3817
Alcalde, M., Farinas, E. T., Arnold, F. H. J. Biomol. Screen. 2004, 9, 141
Alcock, N.W., Bergamini, P., Kemp, T. J., Pringle, P. G. J. Chem. Soc. Chem. Commun.
1987, 235
Algarra, A. G., Grushin, V. V., MacGregor, S. A. Organometallics 2012, 31, 1467
Aluri, S., de Visser, S. P. J. Am. Chem. Soc. 2007, 129, 14846
Amatore, C., Jutand, A. Acc. Chem. Res. 2000, 33, 314
Anderson, G. K., Cross, R. J. Acc. Chem. Res. 2010, 17, 67
Angelelli, F., Chiavarino, B., Crestoni, M. E., Fornarini, S. J. Am. Soc. Mass Spectrom.
2005, 16, 589
Angelici, R. J. Acc. Chem. Res. 1988, 21, 387
Atkins, W. M., Sligar, S. G. J. Am. Chem. Soc. 1987, 109, 3754
252
Atkins, W., Friedman, R. S. Molecular Quantum Mechanics, 3rd
Ed., Oxford University
Press, 1997
Auclair, K., Hu, Z., Little D. M., Ortiz de Montellano, P. R., Groves J. T. J. Am. Chem.
Soc. 2002, 124, 6020
Auclair, K., Moënne-Loccoz, P., Ortiz de Montellano, P. R. J. Am. Chem. Soc. 2001, 123,
4877
Bäckvall, J. E., Åkermark B., Ljunggren, S. O. J. Am.Chem. Soc. 1979, 101, 2411
Bakowies, D., Thiel, W. J. Phys. Chem. 1996, 100, 10580
Balasubramanian, R., Rosenzweig, A. C. Acc. Chem. Res. 2007, 40, 573.
Balcells, D., Raynaud, C., Crabtree, R. H., Eisenstein, O. Chem. Comm. 2008, 744
Barluenga, J., Fernández-Rodríguez, M. A., Aguilar, E. J. Organometal. Chem. 2005,
690, 539
Barnett, S. M., Goldberg, K. I., Mayer, J. M. Nature Chem. 2012, 4, 498
Bartmess, J. E., Georgiadis, R. M. Vacuum 1983, 33, 149
Bartoli, J. F., Brigaud, O., Battioni, P., Mansuy, D. J. Chem. Soc., Chem. Commun. 1991,
440
Bassan, A., Blomberg, M. R. A., Siegbahn, P. E. M., Que Jr., L. Angew. Chem. Int. Ed.
2005, 44, 2939
Bathelt, C. M., Zurek, J., Mulholland, A. J., Harvey, J. N. J. Am. Chem. Soc. 2005, 127,
12900
Becke, A. D. J. Chem. Phys. 1993, 98, 5648
253
Becke, A. D. Phys. Rev. A 1988, 38, 3098
Belcher, J., McLean, K. J., Matthews, S., Woodward, L. S., Fisher, K., Rigby, S. E. J.,
Nelson, D. R., Potts, D., Baynham, M. T., Parker, D. A., Leys, D., Munro, A. W. J. Biol.
Chem. 2014, 289, 6535
Belevich, I., Verkhovsky, M. I., Wikström, M. Nature 2006, 440, 829
Bell, S. R., Groves, J. T. J. Am. Chem. Soc. 2009, 131, 9640
Ben-Daniel, R., de Visser, S. P., Shaik, S., Neumann, R. J. Am. Chem. Soc. 2003, 125,
12116
Bénisvy, L., Halut, S., Donnadieu, B., Tuchagues, J.-P., Chottard, J.-C., Li, Y. Inorg.
Chem. 2006, 45, 2403
Bernal, M. J., Torres, O., Martín, M., Sola, E. J. Am. Chem. Soc. 2013, 135, 19008
Bernasconi, C. F. Chem. Soc. Rev. 1997, 26, 299
Bernasconi, E. J., Baerends, Eur. J. Inorg. Chem. 2008, 1672
Bhowmik, S., Dey, S., Sahoo, D., Rath, S. P. Chem. Eur. J. 2013, 19, 13732
Bhowmik, S., Ghosh, S. K., Layek, S., Verma, H. C., Rath, S. P. Chem. Eur. J. 2012, 18,
13025
Bhowmik, S., Ghosh, S. K., Rath, S. P. Chem. Commun. 2011, 47, 4790
Bhowmik, S., Sil, D., Patra, R., Rath, S. P. J. Chem. Sci. 2011, 123, 827
Bianchini, C., Meli, A. Acc. Chem. Res. 1983, 31, 109
Billig, E., Jamerson, J. D., Pruett, R. L. J. Organomet. Chem. 1980, 192, C49
254
Borovik, S. Acc. Chem. Res. 2005, 38, 54
Borren, E. S., Hill, A. F., Shang, R., Sharma, M., Willis, A. C. J. Am. Chem. Soc. 2013,
135, 4942
Bouaoud, S-E., Braunstein, P., Grandjean, D., Matt, D. Inorg. Chem. 1986, 25, 3765.
Brook, R. L., Borovik, A. S. Inorg. Chem. 2010, 49, 3646
Brookhart, M., Green, M. L. H. J. Organomet. Chem. 1983, 250, 395
Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swaminathan, S., Karplus,
M. J. Comput. Chem. 1983, 4, 187
Brown, J. M., Cooley, N. A. Chem. Rev. 1988, 88, 1031
Bruijnincx, P. C. A., Buurmans, I. L. C., Gosiewska, S., Moelands, M. A. H., Lutz, M.,
Spek, A. L., van Koten, G., Klein Gebbink, R. J. M. Chem. Eur. J. 2008, 14, 1228
Bruijnincx, P. C. A., Buurmans, I. L. C., Gosiewska, S., Moelands, M. A. H., Lutz, M.,
Spek, A. L., van Koten, G., Klein Gebbink, R. J. M. Chem. Eur. J. 2008, 14, 1228
Bruijnincx, P. C. A., van Koten, G., Klein Gebbink, R. J. M. Chem. Soc. Rev. 2008, 37,
2716
Bryant, J. R., Mayer, J. M. J. Am. Chem. Soc. 2003, 125, 10351
Buchowicz, W., Herbaczyńska, B., Jerzykiewicz, L. B., Lis, T., Pasynkiewicz, S.,
Pietrzykowski, A. Inorg. Chem. 2012, 51, 8292
Busca, G., Chapter 1 - Heterogeneous Catalysts, in Heterogeneous Catalytic Materials,
G. Busca, Editor 2014, Elsevier: Amsterdam. p. 1
Calhorda, M. J., Brown J. M., Cooley, N. A. Organometallics, 1991, 10, 1431
255
Caselli, A., Solari, E., Scopelliti, R., Floriani, C. J. Am. Chem. Soc. 2000, 122, 538
Chabinyc, M. L., Craig, S. L., Regan, C. K., Brauman J. I. Science 1983, 279, 1882
Chadwick, R. W., Chuang, L. T., Williams, K. Pestic. Bio-chem. Phys. 1975, 5, 575
Chan, A. S. C., Caroll, W. E., Willis, D. E. J. Mol. Catal. 1983, 19, 377
Che, C.-M., Huang, J.-S. Coord. Chem. Rev. 2002, 231, 151
ChemShell, a Computational Chemistry Shell, see www.chemshell.org, and "QUASI: A
general purpose implementation of the QM/MM approach and its application to problems
in catalysis" Sherwood, P., de Vries, A. H., Guest, M. F., Schreckenbach, G., Catlow, C.
R. A., French, S. A., Sokol, A. A., Bromley, S. T., Thiel, W., Turner, A. J., Billeter, S.,
Terstegen, F., Thiel, S., Kendrick, J., Rogers, S. C., Casci, J., Watson, M., King, F.,
Karlsen, E., Sjøvoll, M., Fahmi, A., Schäfer, A., Ch. Lennartz, J. Mol. Struct.
(Theochem.) 2003, 632, 1
Chen, J. B., Angelici, R. J. Coord. Chem. Rev. 2000, 206, 63
Chen, J., Daniels, L. M., Angelici, R. J. J. Am. Chem. Soc. 1990, 112, 199
Chiavarino, B., Cipollini, R., Crestoni, M. E., Fornarini, S., Lanucara, F., Lapi A., J. Am.
Chem. Soc. 2008, 130, 3208
Chinn, M. S., Heinekey, D. M. J. Am. Chem. Soc. 1990, 112, 5166
Chufán, E. E., Puiu, S. C., Karlin, K. D. Acc. Chem. Res. 2007, 40, 563
Collman, J. P., Boulatov, R., Sunderland, C. J., Fu, L. Chem. Rev. 2004, 104, 561
Collman, J. P., Brauman, J. I., Hampton, P. D., Tanaka, H., Bohle, D. S., Hembre, R. T. J.
Am. Chem. Soc. 1990, 112, 7980
256
Collman, J. P., Kodadek, T., Brauman, J. I. J. Am. Chem. Soc. 1986, 108, 2588
Collman, J. P., Zeng, L., Wang, H. J. H., Lei, A., Brauman, J. I. Eur. J. Org. Chem. 2006,
2707
Comba, P., Maurer, M., Vadivelu, P. Inorg. Chem. 2009, 48, 10389
Comba, P., Rajaraman, G. Inorg. Chem. 2008, 47, 78
Company, A., Feng, Y., Güell, M., Ribas, X., Luis, J. M., Que Jr, L., Costas, M. Chem.
Eur. J. 2009, 15, 3359
Concepcion, J. J., Jurss, J. W., Brennaman, M. K., Hoertz, P. G., Patrocinio, A. O. T., Iha,
N. Y. M., Templeton, J. L., Meyer, T. J. Acc. Chem. Res. 2009, 42, 1954
Cong, Z., Kurahashi, T., Fujii, H. Angew. Chem. Int. Ed. 2011, 50, 9935
Conner, D., Jayaprakash, K. N., Wells, M. B., Manzer, S., Gunnoe, T. B., Boyle, P. D.
Inorg. Chem. 2003, 42, 4759
Conroy-Lewis, F. M., Mole, L., Redhouse, A. D., Litster, S. A., Spencer, J. L. J. Chem.
Soc. Chem. Commun. 1991, 1601
Cooper, H. L. R., Groves, J. T., Arch. Biochem. Biophys. 2011, 507, 111
Corey R.B., Pauling, L. 1953: “Molecular Models of Amino Acids, Peptides, and
Proteins”. Review of Scientific Instruments, Volume 24, Issue 8, 621
Costas, M., Mehn, M. P., Jensen, M. P., Que Jr, L. Chem. Rev. 2004, 104, 939
Crestoni, M.E., Fornarini S., Lanucara F., Warren J.J., Mayer J.M., J. Am. Chem. Soc.
2010, 132, 4336
Crestoni, M.E., Fornarini, S. Inorg. Chem. 2005, 44, 5379
257
Crestoni, M.E., Fornarini, S. Inorg. Chem. 2007, 46, 9018
Crestoni, M.E., Fornarini, S., Lanucara, F. Chem. Eur. J. 2009, 15, 7863
Crevier, T. J., Lovell, S., Mayer, J. M., Rheingold, A. L., Guzei, I. A. J. Am. Chem. Soc.
1983, 120, 6607
Crevier, T. J., Mayer, J. M. J. Am. Chem. Soc. 1983, 120, 5593
Czarnecki, K., Nimri, S., Gross, Z., Proniewicz, L. M., Kincaid, J. R. J. Am. Chem. Soc.
1996, 118, 2929
Dansette, P. M., Bertho, G., Mansuy, D. Biochem. Biophys. Res. Commun. 2005, 338, 450
Das, R. K., Saha, B., Rahaman, S. M. W., Bera, J. K. Chem. Eur. J. 2010, 16, 14459
Dassama, L. M. K., Silakov, A., Krest, C. M., Calixto, J. C., Krebs, C., Bollinger Jr., J.
M., Green, M. T. J. Am. Chem. Soc. 2013, 135, 16758
Davydov, D. R., Bon Hoa, G. H., Peterson, J. A. Biochemistry 1999, 38, 751
Dawson, J. H., Holm, R. H., Trudell, J. R., Barth, G., Linder, R. E., Bunnenberg, E.,
Djerassi, C., Tang, S. C. J. Am. Chem. Soc. 1976, 98, 3707
De Oliveira, T., Chanda, A., Banerjee, D., Shan, X., Mondal, S., Que Jr, L., Bominaar, E.
L., Münck, E., Collins, T. J. Science 2007, 315, 835
de Quadras, L., Bauer, E. B., Stahl, J., Zhuravlev, F., Hampel, F., Gladysz, J. A. New. J.
Chem. 2007, 31, 1594
de Visser, S. P. Angew. Chem. Int. Ed. 2006, 45, 1790
de Visser, S. P. Chem. Eur. J. 2006, 12, 8168
de Visser, S. P. Inorg. Chem. 2006, 45, 9551.
258
de Visser, S. P. J. Am. Chem. Soc. 2006, 128, 15809
de Visser, S. P. J. Am. Chem. Soc. 2006, 128, 9813
de Visser, S. P. J. Am. Chem. Soc. 2010, 132, 1087
de Visser, S. P. J. Biol. Inorg. Chem. 2006, 11, 168
de Visser, S. P. J. Phys. Chem. A 2005, 109, 11050
de Visser, S. P., Kaneti, J., Neumann, R., Shaik, S. J. Org. Chem. 2003, 68, 2903
de Visser, S. P., Kumar, D. (Eds.) Iron-containing enzymes: Versatile catalysts of
hydroxylation reactions in nature. RSC Publishing, Cambridge (UK), 2011
de Visser, S. P., Kumar, D., Cohen, S., Shacham, R., Shaik, S. J. Am. Chem. Soc. 2004,
126, 8362
de Visser, S. P., Latifi, R., Tahsini, L., Nam, W. Chem. Asian J. 2011, 6, 493
de Visser, S. P., Ogliaro F., Harris, N., Shaik S. J. Am. Chem. Soc. 2001, 123, 3037
de Visser, S. P., Ogliaro F., Sharma, P. K., Shaik S. Angew. Chem. Int. Ed. 2002, 41,
1947
de Visser, S. P., Ogliaro, F., Gross, Z., Shaik, S. Chem. Eur. J. 2001, 7, 4954
de Visser, S. P., Ogliaro, F., Harris, N., Shaik, S. J. Am. Chem. Soc. 2001, 123, 3037
de Visser, S. P., Ogliaro, F., Shaik, S. Angew. Chem. Int. Ed. 2002, 41, 1947
de Visser, S. P., Ogliaro, F., Shaik, S. Angew. Chem. Int. Ed. 2001, 40, 2871
259
de Visser, S. P., Ogliaro, F., Sharma, P. K., Shaik, S. Angew. Chem. Int. Ed. 2002, 41,
194
de Visser, S. P., Ogliaro, F., Sharma, P. K., Shaik, S. J. Am. Chem. Soc. 2002, 124, 11809
de Visser, S. P., Oh, K., Han, A.-R., Nam, W. Inorg. Chem. 2007, 46, 4632
de Visser, S. P., Quesne, M. G., Martin. B., Comba, P., Ryde, U. Chem. Commun. 2014,
50, 262
de Visser, S. P., Shaik, S., Sharma, P. K., Kumar, D., Thiel, W. J. Am. Chem. Soc. 2003,
125, 15779
de Visser, S. P., Tahsini, L., Nam, W. Chem. Eur. J. 2009, 15, 5577
de Visser, S. P., Tan, L.-S. J. Am. Chem. Soc. 2008, 130, 12961
de Visser, S.P. Editor 2012, Academic Press. p. 1
de Visser, S.P. Introduction to quantum behaviour – a primer. In Quantum Tunnelling in
Enzyme-Catalysed Reactions, R. K. Alleman and N. S. Scrutton, Eds., RSC Publishing,
Cambridge, 2009, Chapter 2, 18
de Visser, S.P., Coord. Chem. Rev. 2009. 253, 5–6, 754
de Visser, S.P., et al. J. Am. Chem. Soc. 2001. 123, 13, 3037
Dekker, G. P. C. M., Buijs, A., Elsevier, C. J., Vrieze, K., van Leeuwen, P. W. N. M.,
Smeets,W. J. J., Spek, A. L., Wang, Y. F., Stam, C. H. Organometallics, 1992, 11, 1937
Denisov, I. G., Makris, T. M., Sligar, S. G., Schlichting, I. Chem. Rev. 2005, 105, 2253
Denmark, S. E., Fu, J. Chem. Rev. 2003, 103, 2763.
Dey, R. K. Hocking, P. L. Larsen, A. S. Borovik, B. Hedman, K. O. Hodgson, E. I.
Solomon, J. Am. Chem. Soc. 2006, 128, 9825
260
Diver, S. T. Coord. Chem. Rev. 2007, 251, 671
Dolinsky, T. J., Czodrowski, P., Li, H., Nielsen, J. E., Jensen, J. H., Klebe, G., Baker, N.
A. Nucl. Acids Res. 2007, 35, 522
Dong, L., Duckett, S. B., Ohman K. F., Jones, W. D. J. Am. Chem. Soc. 1992, 114, 151
Dowers, T. S., Rock, D. A., Jones, J. P. J. Am. Chem. Soc. 2004, 126, 8868
Duan, L., Bozoglian, F., Mandal, S., Stewart, B., Privalov, T., Llobet, A., Sun, L. Nature
Chem. 2011, 4, 418
Dunford, H. B., Heme Peroxidases, Wiley-VCH, New York, 1999
E. MacBeth, R. Gupta, K. R. Mitchell-Koch, V. G. Young Jr., G. H. Lushington, W. H.
Thompson, M. P. Hendrich, A. S. Borovik, J. Am. Chem. Soc. 2004, 126, 2556
E. O'Reilly, V. Koehler, S. Flitsch and N. Turner, Chem. Commun. 2011, 47, 2490,
England, J., Guo, Y., Van Keuvelen, K.M., Cranswick, M. A., Rohde, G. T., Bominaar,
E. L., Münck, E., Que Jr, L. J. Am. Chem. Soc. 2011, 133, 11880
Enthaler, S. ACS Catalysis, 2013. 3, 2, 150
Evans, D. R., Mathur, R. S., Heerwegh, K., Reed, C. A., Xie, Z. Angew. Chem., Int. Ed.
Engl. 1997, 36, 1335
Ferguson-Miller, S., Babcock, G. T. 1996, 96, 2889
Fernández, I., Frenking, G. Chem. Commun. 2006, 5030.
Fioroni, M., T. Dworeck, and F. Rodríguez-Ropero, Theoretical Considerations and
Computational Tools, in ß-barrel Channel Proteins as Tools in Nanotechnology 2014,
Springer Netherlands. p. 69
261
Fitton, P., Rick, E. A. J. Organomet. Chem. 1971, 28, 287
Fleming, I. Frontier Orbitals and Organic Chemical Reactions, Wiley, New York 1976,
161.
Flood, T.C., Jensen, J.E., Statler, J.A. J. Am. Chem. Soc. 1981, 103, 4410
Forkert, P.-G., Lee, R. P. Toxicol. Appl. Pharmacol. 1997, 146, 245
Fox, B. G., Lyle, K. S., Rogge, C. E. Acc. Chem. Res. 2004, 37, 421
Franke, M. Wolak and R. van Eldik, Chem. Eur. J. 2009, 15, 10182
Frisch, J. et al. Gaussian-03, Revision C.02, Wallingford, CT, 2003
Frisch, J. et al. Gaussian-09, Revision A.02, Wallingford, CT, 2009
Frisch, M. J. Gaussian 09, revision C.01, Gaussian, Inc.: Wallingford, CT, 2010
Frisch, M. J. Gaussian 03, revision D.01, Gaussian, Inc., Wallingford, CT, 2004
Fürstner, A. Angew. Chem. Int. Ed. 2000, 39, 3012
Fukuzumi, S., Morimoto, Y., Kotani, H., Naumov, P., Lee, Y.-M., Nam, W. Nature
Chem. 2010, 2, 756
Fukuzumi, S., Ohkubo, K. Coord. Chem. Rev. 2010, 254, 372
Galonić, P., Barr, E. W., Walsh, C. T., Bollinger Jr, J. M., Krebs, C. Nat. Chem. Biol.
2007, 3, 113
García Alonso, F. J., Llamazares, A., Riera, V., Vivanco, M., Diaz M. R., García Granda,
S. J. Chem. Soc. Chem. Commun. 1991, 1058
Garrou, P. E. Chem. Rev. 1985, 85, 171
262
Gelb, M. H., Heimbrook, D. C., Mälkönen, P., Sligar, S. G. Biochemistry 1982, 21, 370
Ghosh, A., Acc. Chem. Res. 1983, 31, 189
Ghosh, A., Taylor, P. R. Curr. Opin. Chem. Biol. 2003, 7, 113,
Ghosh, S. K., Patra, R., Rath, S. P. Inorg. Chem. 2008, 47, 10196
Ghosh, S. K., Patra, R., Rath, S. P. Inorg. Chem. 2010, 49, 3449
Ghosh, S. K., Patra, R., Rath, S. P. Inorg. Chim. Acta. 2010, 363, 2791
Goel, A. B. Inorg. Chim. Acta, 2010, 84, L25.
Goertz, W., Kamer, P. C. J., van Leeuwen, P. W. N. M., Vogt, D. Chem, Eur. J. 2001,
7,1614
Gonzalez-Diaz, H., et al. Current Topics in Medicinal Chemistry, 2013. 13,14, 1713
Goodson, F. E., Wallow, T. I., Novak, B. M. J. Am. Chem. Soc. 1997, 119, 12441
Green, M. T. J. Am. Chem. Soc. 1999, 121, 7939
Green, M. T. J. Am. Chem. Soc. 2006, 128, 1902
Green, M. T., Dawson, J. H., Gray, H. B. Science 2004, 304, 1653
Grimme, S. J. Comput. Chem. 2006, 27, 1787
Grimme, S., Antony, J., Ehrlich, S., Krieg, H. J. Chem. Phys. 2010, 132, 154104
Grogan, G., Curr. Opin. Chem. Biol. 2011, 15, 241
Gross, Z. J. Biol. Inorg. Chem. 1996, 1, 368
Gross, Z., Nimri, S. Inorg. Chem. 1994, 33, 1731
Groves, J. T. Gross, Z., Stern, M. K. Inorg. Chem. 1994, 33, 5065
263
Groves, J. T. Models and Mechanisms of Cytochrome P450 Action. In Cytochrome P450:
Structure, Mechanism and Biochemistry. Ortiz de Montellano P.R., (Ed.), 3rd Ed.,
Kluwer Academic/Plenum Publishers, New York, 2005, Chapter 1, pp 1
Groves, J. T. Proc. Natl. Acad. Sci. USA 2003, 100, 3569
Groves, J. T., Ahn, K.-H. Inorg. Chem. 1987, 26, 3831
Groves, J. T., Ahn, K.-H., Quinn, R. J. Am. Chem. Soc. 1988, 110, 4217
Groves, J. T., Avaria-Neisser, G. E., Fish, K. M., Imachi, M., Kuczkowski, R. L. J. Am.
Chem. Soc. 1986, 108, 3837
Groves, J. T., Bonchio, M., Carofiglio, T., Shalyaev, K. J. Am. Chem. Soc. 1996, 118,
8961
Groves, J. T., Gross, Z., Stern, M. K. Inorg. Chem. 1994, 33, 5065
Groves, J. T., McClusky, G. A. J. Am. Chem. Soc. 1976, 98, 859.
Groves, J. T., Myers, R. S. J. Am. Chem. Soc. 1983, 105, 5791
Groves, J. T., Quinn, R. J. Am. Chem. Soc. 1985, 107, 5790
Groves, J. T., Roman, J. C. J. Am. Chem. Soc. 1995, 117, 5594
Groves, J. T., Shalyaev, K., Lee, J. In The Porphyrin Handbook, Kadish, K. M., Smith, K.
M., Guilard, R. Eds., Academic Press: New York, 2000, Vol. 4, Chapter 27, pp. 17
Groves, J. T., Subramanian, D. V. J. Am. Chem. Soc. 2010, 106, 2177
Groves, J. T., Watanabe, Y. J. Am. Chem. Soc. 1986, 108, 507
Groves, T., Stern, M. K., J. Am. Chem. Soc. 1988, 110, 8628
264
Guan, X., Fisher, M. B., Lang, D. H., Zheng, Y. M., Koop, D. R., Rettie, A. E. Chem.
Biol. Interact. 1983, 110, 103
Guengerich, F. P. Arch. Biochem. Biophys. 2003, 409, 59
Guengerich, F. P. Chem. Res. Toxicol. 2001, 14, 611
Guengerich, F. P. Chem. Res. Toxicol. 2008, 21, 70
Guengerich, F. P., Kim, D. H. Chem. Res. Toxicol. 1991, 4, 413
Guengerich, P. Chem. Res. Toxicol. 2001, 14, 611
Gütlich, P., Hauser, A., Spiering, H. Angew. Chem. Int. Ed. Engl. 1994, 33, 2024
Gunes, A., Bilir, E., Zengil, H., Babaoglu, M. O., Bozkurt, A., Yasar, U. Pharmacol.
Toxicol. 2007, 100, 383
Gupta, A., Gupta, G. P. J. Phys. Condens. Matter 1994, 6, 11071
Gupta, R., Lacey, D. C., Bominaar, E. L., Borovik, A. S., Hendrich, M. P. J. Am. Chem.
Soc. 2012, 134, 9775
Hagen, W. R., van den Berg., van Dongen W. A. M., Reijerse W. M. A. M., E. J., van
Kan, P. J. M. J. Chem. Soc. Farad. Trans. 1983, 94, 2969
Handy, N. C., Cohen, A. J. Mol. Phys. 2001, 99, 403
Harding, D. J., Kerpal, C., Meijer, G., Fielicke, A. J. Phys. Chem. Lett. 2013, 4, 892
Harrison, J.G., 1987. “Electron affinities in the self-interaction-correlated local spin
density approximation”. J. Chem. Phys, 86, 2849.
Hay, P. J., Wadt, W. R. J. Chem. Phys. 1985, 82, 270
Hehre, W. J., Ditchfield, K., Pople, J. A. J. Chem. Phys. 1972, 56, 2257
265
Hellmann, H. J. Chem. Phys., 1935, 3, 61
Hessenauer-Ilicheva, N., Franke, A., Meyer, D., Woggon, W.-D. van Eldik, R. J. Am.
Chem. Soc. 2007, 129, 12473
Hessenauer-Ilicheva, N., Franke, A., Meyer, D., Woggon, W.-D., van Eldik, R. Chem.
Eur. J. 2009, 15, 2941
Hessenauer-Ilicheva, N., Franke, A., Meyer, D., Woggon, W.-D., van Eldik, R. J. Am.
Chem. Soc. 2007, 129, 12473
Hessenauer-Ilicheva, N., Franke, A., Wolak, M., Higuchi, T., van Eldik, R. Chem. Eur. J.
2009, 15, 12447
Heyes, D. J., Sakuma, M., de Visser, S. P., Scrutton, N. S. J. Biol. Chem. 2009, 284, 3762
Hickey, C. E., Maitlis, P. M. J. Chem. Soc. Chem. Commun. 2010, 1609
Hirao, H., Kumar, D., Que Jr, L., Shaik, S. J. Am. Chem. Soc. 2006, 128, 8590
Hirao, H., Kumar, D., Thiel, W., Shaik, S. J. Am. Chem. Soc. 2005, 127, 13007
Hirao, H., Que Jr, L., Nam, W., Shaik, S. Chem. Eur. J. 2008, 14, 1740
Hoe, W.-M., Cohen, A., Handy, N. C. Chem. Phys. Lett. 2001, 341, 319
Hoffart, L. M., Barr, E. W., Guyer, R. B., Bollinger Jr, J. M., Krebs C. Proc. Natl. Acad.
Sci. USA 2006, 103, 14738
Hohenberg, P., Kohn, W., Phys. Rev. B, 1964, 136, 864
Holm, R., Kennepohl, P., Solomon, E. I., 1996, Chemistry Reviews, 96, 2239
Hückel, Z. Physik, 1931, 70, 204.
266
Hull, J. F., Balcells, D., Sauer, E. L. O., Raynaud, C., Brudvig, G. W., Crabtree, R. H.,
Eisenstein, O. J. Am. Chem. Soc. 2010, 132, 7605
Hull, J. F., Sauer, E. L. O., Incarvito, C. D., Faller, J. W., Brudvig, G. W., Crabtree, R. H.
Inorg. Chem. 2009, 48, 488
Hurst, J. K. Coord. Chem. Rev. 2005, 249, 313
Huynh, M. H. V., White, P. S., Carter, C. A., Meyer, T. J. Angew. Chem., Int. Ed. 2001,
40, 3037
Ikariya, T., Murata, K., Noyori, R. Org. Biomol. Chem. 2006, 4, 393
Inuki, T., Kojima, T. J. Org. Chem. 1967, 32, 869
Ishihara, K., Kurihara, H., Matsumoto, M., Yamamoto, H. J. Am. Chem. Soc. 1983, 120,
6920
Ishihara, K., Ohara, S., Yamamoto, H. Science 2000, 290, 1140
Isobe, H., Yamanaka, S., Okumura, M., Yamaguchi, K., Shimada, J. J. Phys. Chem. B
2011, 115, 10730
Itoh, S. Curr. Opin. Chem. Biol. 2006, 10, 115
Jackson, T. A., Rohde, J.-U., Seo, M. S., Sastri, C. V., DeHont, R., Stubna, A., Ohta, T.,
Kitagawa, T., Münck, E., Nam, W., Que Jr, L. J. Am. Chem. Soc. 2008, 130, 12394
Jaguar 7.6, Schrödinger LLC, New York NY, 2007
Jaguar 7.7, version 7.9, Schrodinger, LLC, New York, NY, 2011
Jaguar 7.8, Schrödinger LLC, New York NY, 2010
267
Jaguar, version 7.9, Schrödinger, LLC, New York, NY, 2011
James, B. R., Mahajan, D. Can. J. Chem. 1979, 57, 180
Jensen, F. Introduction to Computational Chemistry, 2nd
Ed., John Wiley & Sons, Ltd.,
Chichester, 2007
Jensen, H. Molecular Modeling Basics. CRC Press, Boca Raton, 2010
Ji, L., Franke, A., Brindell, M., Oszajca, M., Zahl, A., van Eldik, R. Chem. Eur. J. 2014,
20, 2347.
Ji, L., Schüürmann, G. Angew. Chem. Int. Ed. 2013, 52, 744
Ji, L., Zhang, J., Liu, W., de Visser, S. P. Chem. Asian J. 2014, 9, 1175
Johansson, J., Blomberg, M. R. A., Siegbahn, P. E. M. J. Phys. Chem. C 2007, 111,
12397
Johnston, J. B., Kells, P. M., Podust, L. M., Ortiz de Montellano, P. R. Proc. Natl. Acad.
Sci. USA. 2009, 106, 20687
Kadish, K. M., Smith, K. M., Guilard, R. Eds. Handbook of Porphyrin Science, World
Scientific Publishing Co.: New Jersey, 2010
Kahn, O. Molecular Magnetism, VCH Publishers: Weinheim, 1993, 2
Kaizer, J., Klinker, E. J., Oh, N.Y., Rohde, J.-U., Song, W. J., Stubna, A., Kim, J., Münck
E., Nam, W., Que Jr, L. J. Am. Chem. Soc. 2004, 126, 472
Kamachi, T., Shiota, Y., Ohta, T., Yoshizawa, K. Bull. Chem. Soc. Jpn. 2003, 76, 721
Kamachi, T., Yoshizawa, K. J. Am. Chem. Soc. 2003, 125, 4652
268
Kaminsky, L. S., Fasco, M. J., Guengerich, F. P. J. Biol. Chem. 1980, 255, 85
Kaneda, K., Sano, K., Teranishi, S. Chem. Lett. 1979, 82
Karamzadeh, B., Kumar, D., Sastry, G. N., de Visser, S. P. J. Phys. Chem. A 2010, 114,
13234
Kataoka, Y., Shibahara, A., Yamagata, T., Tani, K. Organometallics 2001, 20, 2431
Kelly, S. L., Lamb, D. C., Kelly, D. E. FEBS. Lett. 1997, 412, 233
Kenneth S. K. 1987. Introductory Nuclear Physics. Wiley. ISBN 978-0-471-80553-3.
Kikukawa, K., Takagi, M., Matsuda, T. Bull. Chem. Soc. Japan, 1979, 52, 1493.
Kitchen, J. A., Brooker, S. Coord. Chem. Rev. 2008, 252, 2072
Kitchen, J. A., White, N. G., Jameson, G. N. L., Tallon, J. L., Brooker, S. Inorg. Chem.
2011, 50, 4586
Kohn, W., Sham, L. J. Phys. Rev. A, 1965, 140, 1133
Kong, K.-C., Cheng, C.-H. J. Am. Chem. Soc. 1991, 113, 6313
Korendovych, I. V., Kryatov, S. V., Rybak-Akimova, E. V. Acc. Chem. Res. 2007, 40,
510
Kovaleva, G., Neibergaul, M.B., Chakrabarty, S., Lipscomb, J. D., Acc. Chem. Res. 2007,
40, 475
Kratsch, J., Roesky, P.W. Angewandte Chemie International Edition, 2014. 53, 2, 376
269
Krebs, C., Dassama, L. M. K., Matthews, M. L., Jiang, W., Price, J. C., Korboukh, V., Li,
N., Bollinger Jr, J. M. Coord. Chem. Rev. 2013, 257, 234
Kryatov, S. V., Rybak-Akimova, E. V., Schindler, S. Chem. Rev. 2005, 105, 2175
Kumar, A., Goldberg, I., Botoshansky, M., Buchman, Y., Gross, Z. J. Am. Chem. Soc.
2010, 132, 15233
Kumar, D, Latifi, R., Kumar, S., Rybak-Akimova, E. V., Sainna, M. A., de Visser, S. P.
Inorg. Chem. 2013, 52, 7968
Kumar, D., de Visser, S. P., Shaik S. Chem. Eur. J. 2005, 11, 2825
Kumar, D., de Visser, S. P., Shaik S. J. Am. Chem. Soc. 2003, 125, 13024
Kumar, D., de Visser, S. P., Shaik, S., J. Am. Chem. Soc. 2004, 126, 5072
Kumar, D., de Visser, S. P., Sharma, P. K., Cohen, S., Shaik, S. J. Am. Chem. Soc. 2004,
126, 1907
Kumar, D., de Visser, S. P., Sharma, P. K., Derat, E., Shaik, S. J. Biol. Inorg. Chem.
2005, 10, 181
Kumar, D., Derat, E., Khenkin, A. M., Neumann, R., Shaik, S. J. Am. Chem. Soc. 2005,
127, 17712
Kumar, D., Hirao, H., Que Jr, L., Shaik, S. J. Am. Chem. Soc. 2005, 127, 8026
Kumar, D., Karamzadeh, B., Sastry, G. N., de Visser, S. P. J. Am. Chem. Soc. 2010, 132,
7656
Kumar, D., Sastry, G. N, de Visser, S. P., J. Phys. Chem. B 2012, 116, 718
Kumar, D., Sastry, G. N., de Visser, S. P. Chem. Eur. J. 2011, 17, 6196
270
Kumar, D., Tahsini, L., de Visser, S.P., Kang, H. Y., Kim, S. J., Nam, W. J. Phys. Chem.
A 2009, 113, 11713
Kumar, D., Thiel, W., de Visser, S. P. J. Am. Chem. Soc. 2011, 133, 3869
Kumar, S., Faponle, A. S., Barman, P., Vardhaman, A. K., Sastri, C. V., Kumar. D., de
Visser, S. P. J. Am. Chem. Soc. 2014, in press, DOI 10.1021/ja508403w.
Kurahashi, T., Kikuchi, A., Tosha, T., Shiro, Y., Kitagawa, T., Fujii, H. Inorg. Chem.
2008, 47, 1674
Lafite, P., André, F., Zeldin, D. C., Dansette, P. M., Mansuy, D. Biochemistry 2007, 46,
10237
Lansky, D. E., Goldberg, D. P., Inorg. Chem. 2006, 45, 5119
Lanucara, F., Chiavarino, B., Crestoni, M.E., Scuderi, D., Sinha, R.K., Maitre P.,
Fornarini S., Inorg. Chem. 2011, 50, 4445
Lanucara, F., Crestoni, M. E. Chem. Eur. J. 2011, 17, 12092
Latifi, R., Sainna, M. A., Rybak-Akimova, E. V., de Visser, S. P. Chem. Eur. J. 2013,
19, 4058
Latifi, R., Bagherzadeh, L., de Visser, S. P. Chem. Eur. J. 2009, 15, 6651
Latifi, R., J. S. Valentine, W. Nam, S. P. de Visser, Chem. Commun. 2012, 48, 3491
Latifi, R., Tahsini L., B. Karamzadeh, N. Safari, W. Nam, S. P. de Visser, Arch. Biochem.
Biophys. 2011, 507, 4
Latifi, R., Tahsini L., Kumar, D., Sastry, G. N., Nam, W., de Visser, S. P. Chem.
Commun. 2011, 47, 10674
271
Lee, C., Yang, W., Parr, R. G. Phys. Rev. B 1988, 37, 785
Lee, R. P., Parkinson, A., Forkert, P.-G. Drug Metabol. Disp. 1983, 26, 60
Leeladee, P., Baglia, R. A., Prokop, K. A., Latifi, R., de Visser, S. P., Goldberg, D. P. J.
Am. Chem. Soc. 2012, 134, 10397
Leeladee, P., Goldberg, D. P. Inorg. Chem. 2010, 49, 3083
Leithauser, M. T., Liem, A., Stewart, B. C., Miller, E. C., Miller, J. A. Carcinogenesis
1990, 11, 463
Lewis, J.C. ACS Catal 2013, 3, 12, 2954
Li, J., Lord, R. L., Noll, B. C., Baik, M.-H., Schulz, C. E., Scheidt, W. R. Angew. Chem.
Int. Ed. 2008, 47, 10144
Lias, S. G., Bartmess, J. E., Liebman, J. F., Holmes, J. L., Levin, R. D., Mallard, W. G.,
In Ion Energetics Data in NIST Chemistry WebBook, Linstrom, P. J., Mallard, W. G.
(Eds.), NIST Standard Reference Database Number 69, National Institute of Standards
and Technology: Gaithersburg, MD, 20899 (http://webbook.nist.gov).
Lide, D. R. (Ed) Handbook of Chemistry and Physics, CRC Press, Boca Raton, New
York, 1996
Lin, Y. A., Davis, B. G. Beilstein J. Org. Chem. 2010, 6, 1219
Linde, C., Åkermark, B., Norrby, P.-O., Svensson, M. J. Am. Chem. Soc. 1999, 121, 5083
Ling, Y., Zhang, Y. J. Am. Chem. Soc. 2009, 131, 6386
272
Loch, J. M., Potter, J., Bachmann, K. A. Pharmacol. 1995, 50, 146
Lonsdale, R., Harvey, J. N., Mulholland, A. J. J. Phys. Chem. Lett. 2010, 1, 3232
Lonsdale, R., Oláh, J., Mulholland, A. J., Harvey, J. N. J. Am. Chem. Soc. 2011, 133,
15464
Love, J. A., Sanford, M. S., Day, M. W., Grubbs, R. H. J. Am. Chem. Soc. 2003, 125,
10103
Lowe, John P. Quantum Chemistry, Academic Press: New York, 1978, pp 315
Lu Y, Yeung N, Sieracki N, Marshall N., M. Nature 2009, 460,855
Lundquist, E. G., Folting, K., Huffman, J.C., Caulton, K.G. Organometallics 1990, 7,
2254.
Luo, X.-L., Crabtree, R. H. J. Am. Chem. Soc. 1989, 111, 2527
MacGregor, S. A., McKay, D., Panetier, J. A., Whittlesey, M. K. Dalton Trans. 2013, 42,
7386
Mader, A., Manner, V. W., Markle, T. F., Wu, A., Franz, J. A., Mayer, J. M. J. Am.
Chem. Soc. 2009, 131, 4335
Maestri, A. G., Cherry, K. S., Toboni, J. J., Brown, S. N. J. Am. Chem. Soc. 2001, 123,
7459
Major, A., Horváth, I. T., Pino, P. J. Mol. Catal. 1988, 45, 275
Man, W.-L., Lam, W. W. Y., Lau, T.-C. Acc. Chem. Res. 2014, 47, 427
273
Manner, V. W., Lindsay, A. D., Mader, E. A., Harvey, J. N., Mayer, J. M., Chem. Sci.
2012, 3, 230
Marcone, J. E., Moloy, K. G. J. Am. Chem. Soc. 1983, 120, 8527
Marshall G., Hendrickson C.L., Jackson G.S., Mass Spec. Rev. 1983, 17, 1G. Marshall, C.
L. Hendrickson and G. S. Jackson, Mass Spec. Rev. 1983, 1
Martinho, M., Banse, F., Bartoli, J.-F., Mattioli, T. A., Battioni, P., Horner, O., Bourcier,
S., Girerd, J.-J. Inorg. Chem. 2005, 44, 9592
Mas-Balleste, R., Que Jr, L. J. Am. Chem. Soc. 2007, 129, 15964
Maseras, F., Lledós, A. Computational Modeling of Homogeneous Catalysis, Kluwer
Academic Publishers: Dordrecht, The Netherlands, 2002.
Matsuo, T., Mayer, J. M. Inorg. Chem. 2005, 44, 2150
Mawby, R.J., Basolo, F., Pearson, R.G. J. Am. Chem. Soc. 1964, 86, 5043
Mayer, J. M. Acc. Chem. Res. 1983, 31, 441
Mayer, M. Annu. Rev. Phys. Chem. 2004, 55, 363
McGown, A. J., Kerber, W. D., Fujii, H., Goldberg, D. P. J. Am. Chem. Soc. 2009, 131,
8040
McLean, K. J., Munro, A. W. Drug Metabol. Rev. 2008, 40, 427
Meot-Ner, M., In Gas Phase Ion Chemistry. M. T. Bowers (Ed.), Academic Press, New
York, 1979, Vol. 1
Meunier, B., de Visser, S. P., Shaik, S. Chem. Rev. 2004, 104, 3947
Meyer, T. J., Huynh,M. H. V. Inorg. Chem. 2003, 42, 8140
274
Miller, R. L., Wolczanski, P. T., Rheingold, A. L. J. Am. Chem. Soc. 1993, 115, 10422
Millikan, R.A., Physical Review. 1911, 32,349
Mirbach, M. F. J. Metalorg. Chem. 2010, 265, 205.
Moënne-Loccoz, P., Krebs, C., Herlihy, K., Edmondson, D. E., Theil, E. C., Huynh, B.
H., Loehr, T. M. Biochemistry 1999, 38, 5290
Mole, L., Spencer, J. L., Carr, N., Orpen, A. G. Organometallics, 1991, 10, 49
Molecular Quantum Mechanics Parts I and II: An Introduction to QUANTUM
CHEMISTRY (Volume 1), Atkins, P.W. Oxford University Press, 1977
Møller, C., Plesset, M. S. Phys. Rev., 1934, 46, 0618
Monsaert, S., Lozano Vila, A., Drozdzak, R., van der Voort, P., Verpoort, F. Chem. Soc.
Rev. 2009, 38, 3360
Munro, A. W., Girvan, H. M., McLean, K. J. Nat. Prod. Rep. 2007, 24, 585
Murphy, C. Biotechnology Letters, 2014, 1
Murray, L. J., Lippard, S. J. Acc. Chem. Res. 2007, 40, 466
Mutter, S. T., Platts, J. A. J. Phys. Chem. A 2011, 115, 11293
Nagata, K., Liberato, D. J., Gillette, J. R., Sasame, H. A. Drug Metab. Dispos. 1986, 14,
559
Nakamura, K., Ikezaki, A., Ohgo, Y., Ikeue, T., Neya, S., Nakamura, M. Inorg. Chem.
2008, 47, 10299
275
Nakamura, M., Ohgo, Y., Ikezaki, A., Electronic and Magnetic Structures of Iron
Porphyrin Complexes, In Handbook of Porphyrin Science, Kadish, K.M., Smith, K.M.,
Guilard, R. (eds.), World Scientific, 2010, 1
Nam, W. Acc. Chem. Res. 2007, 40, 522
Nam, W., et al. Am. Chem. Soc. 2000. 122, 28, 6641
Nam, W., Ryu, Y.O., Song, W. J. J. Biol. Inorg. Chem. 2004, 9, 654
Neese, ORCA – an ab initio, DFT and semiempirical SCF-MO package, Bonn, Germany,
Version 2.7-00, 2009
Nehru, K., Kim, S. J., Kim, I. Y., Seo, M. S., Kim, Y., Kim, S.-J., Kim, J., and Nam, W.
Chem. Commun. 2007, 4623
Nelson, D. R. Human Genomics 2009, 4, 59
Nelson, D.R., Methods Mol. Biol. 2006, 320, 1
Neu, H. M., Quesne, M.G., Yang, T., Prokop-Prigge, K.,A., Lancaster K.,M., Donohoe J.,
DeBeer, S., de Visser, S. P., Goldberg, D. P. Chem. Eur. J. 2014, 20, 14584
Nibbering, N. M. M. Mass Spec. Rev. 2006, 25, 962
Nihei, M., Shiga, T., Maeda, Y., Oshio, H. Coord. Chem. Rev. 2007, 251, 2606
Noack, K.,Calderazzo, F. J. Organometal. Chem. 1967, 10, 101
Nordlund, P., Reichard, P. Annu. Rev. Biochem. 2006, 75, 681
Nyman, G., Int. J. of Quan. Chem., 2014. 114, 1183
O’Brien, J. R., Schuller, D. J., Yang, V. S., Dillard, B. D., Lanzilotta W. N. Biochemistry
2003, 42, 5547
276
O’Connor, C. J. Prog. Inorg. Chem. 1982, 29, 203.
Ogliaro, F., de Visser, S. P., Cohen, S., Kaneti, J., Shaik, S. ChemBioChem. 2001, 11, 848
Ogliaro, F., Harris, N., Cohen, S., Filatov, M., de Visser, S. P., Shaik, S. J. Am. Chem.
Soc. 2000, 122, 8977
Ohno, T., Suzuki, N., Dokoh, T., Urano, Y., Kikuchi, K., Hirobe, M., Higuchi, T.,
Nagano, T. J. Inorg. Biochem. 2000, 82, 123,
Omura, T., Sato, R. J. Biol. Chem. 1962, 237, 1375
O'Reilly, E., Koehler, V., Flitsch, S., Turner, N. Chem. Commun. 2011, 47, 2490
Ortiz de Montellano, P. R. Chem. Rev. 2010, 110, 932
Ortiz de Montellano, P. R. Ed. Cytochrome P450: Structure, Mechanism and
Biochemistry. 3rd ed., Kluwer Academic/Plenum Publishers, New York, 2004
Ortiz de Montellano, P. R., Fruetel, J. A., Collins, J. R., Camper, D. L., Loew, G. H. J.
Am. Chem. Soc. 1991, 113, 3195
Ostović, D., Bruice, T.C., J. Am. Chem. Soc. 1989, 111, 6511
Pan, Z., Zhang, R., Newcomb, M. J. of Inorg. Biochem, 2006,100, 524
Parsell, T. H., Behan, R. K., Green, M. T., Hendrich, M. P., Borovik, A. S. J. Am. Chem.
Soc. 2006, 128, 8728
Parsell, T. H., Yang, M.-Y., Borovik, A. S. J. Am. Chem. Soc. 2009, 131, 2762
Patra, S. K., Bera, J. K. Organometallics 2006, 25, 6054
Patra, S. K., Sadhukhan, N., Bera, J. K. Inorg. Chem. 2006, 45, 4007
Perdew, J. P. Phys. Rev. B 1986, 33, 8822
277
Perdew, J. P., Burke, K., Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865
Perdew, J. P., Wang, Y. Phys. Rev. B 1992, 45, 13244
Peters, J. C., Odom, A. L., Cummins, C. C. Chem. Commun. 1997, 1995
Petrik, I. D., Liu, J., Lu, Y. Curr Opin Chem Biol 2014, 19, 0, 67
Pfaff F., Kundu S., Risch M., Pandian S., Heims F., Pryjomska-Ray I., Haack P.,
Metzinger R., Bill E., Dau H., Comba P., Ray K., Angew. Chem. Int. Ed. 2011, 50, 1711
Posner, G. H., O’Neill, P. M. Acc. Chem. Res. 2004, 37, 397
Poulos, T. L. Peroxidases and Cytochrome P450. In The Porphyrin Handbook, Kadish,
K. M., Smith, K. M., Guilard, R. Eds., Academic Press, San Diego, 2000: Vol. 4, Chapter
32, 189
Poulos, T. L., Finzel, B. C., Gunsalus, I. C., Wagner, G. C., Kraut, J. J. Biol. Chem. 1985,
260, 6122
Poulos, T. L., Finzel, B. C., Howard, A. J. Biochemistry 1986, 25, 5314.
Prat, I., Mathieson, J. S., Güell, M., Ribas, X., Luis, J. M., Cronin, L., Costas, M. Nature
Chem. 2011, 3, 788
Price, J. C., Barr, E. W., Tirupati, B., Bollinger Jr, J. M., Krebs, C. Biochemistry 2003,
42, 7497
Prokop, K. A., de Visser, S. P., Goldberg, D. P. Angew. Chem. Int. Ed. 2010, 49, 5091
Prokop, K. A., Neu, H. M., de Visser, S. P., Goldberg, D. P. J. Am. Chem. Soc. 2011, 133,
15874
278
Quesne, M. G., Latifi, R., Gonzalez-Ovalle, L. E., Kumar, D., de Visser, S. P. Chem. Eur.
J. 2014, 20, 435
Quiñinero, D., Musaev, D. G., Morokuma, K. Inorg. Chem. 2003, 42, 8449
Quintanar, L., Stoj, C., Taylor, A. B., Hart, P. J., Kosman, D. J., Solomon, E. I. Acc.
Chem. Res. 2007, 40, 445
Radoń, M., Broclawic, E., Pierloot, K. J. Chem. Theory Comput. 2011, 7, 898
Rath, S. P., Olmstead, M. M., Balch, A. L. Inorg. Chem. 2006, 45, 6083
Raven, L. Nat. Prod. Rep. 2003, 20, 367
Reed, C. A., Mashiko, T., Bentley, S. P., Kastner, M. E., Scheidt, W. R., Spartalian, K.,
Lang, G. J. Am. Chem. Soc. 1979, 101, 2948
Reetz, M. T., Kyung, S. H., Bolm, C., Zierke, T. Chem. & Ind. (London) 1986, 824
Reetz, M.T., Angewandte Chemie International Edition, 2014. 53, 33, 8562
Reiher, M., Salomon O., Hess, B. A. Theor. Chem. Acc., 2001, 107, 48
Rettie, A. E., Boberg, M., Rettenmeier, A. W., Baillie, T. A. J. Biol. Chem. 1988, 263,
13733
Rettie, A. E., Rettenmeier, A. W., Howald, W. N., Baillie, T. A. Science 1987, 235, 890
Reynald, R. L., Sansen, S., Stout, C. D., Johnson, E. F. J. Biol. Chem. 2012, 287, 44581
Riley, K. E., Op’t Holt, B. T., Merz, K. M. J. J. Chem. Theory Comput., 2007, 3, 407
Rittle, J., Green, M. T. Science 2010, 330, 933
Roberts, K. M., Jones, J. P. Chem. Eur. J. 2010, 16, 8096
279
Rohde, J-U., In J-H., Lim, M. H., Brennessel, W. W., Bukowski, M. R., Stubna, A.,
Münck, E., Nam, W., Que Jr, L. Science 2003, 299, 1037
Roithová, J., Schröder, D. Chem. Rev. 2010, 110, 1170
Romain, S., Vigara, L., Llobet, A. Acc. Chem. Res. 2009, 42, 1944
Rude, M. A., Baron, T. S., Brubaker, S., Alibhai, M., Del Cardayre, S. B., Schirmer, A.
Appl. Env. Microbiol. 2011, 77, 1718
Ruettinger, R. T., Fulco, A. J. J. Biol. Chem. 1981, 256, 5728
Ruettinger, R. T., Wen, L. P., Fulco, A. J. J. Biol. Chem. 1989, 264, 10987
Sadeque, A. J. M., Fisher, M. B., Korzekwa, K. R., Gonzalez, F. J., Rettie, A. E. J.
Pharmacol. Exp. Therap. 1997, 283, 698
SAINT+, 6.02 ed., Bruker AXS, Madison, WI, 1999
Sakakura, T. J. Organometal. Chem. 2010, 267, 171
Sala, X., Romero, I., Rodríguez, M., Escriche, L., Llobet, A. Angew. Chem. Int. Ed. 2009,
48, 2842
Saltzman, H., Sharefkin, J.G. Organic Syntheses. Wiley. New York. 1973, Collect. Vol. 5,
658
Sastri, V., Lee, L., Oh, K., Lee, Y. J., Lee, J., Jackson, T. A., Ray, K., Hirao, H., Shin,
W., Halfen, J. A., Kim, J., Que Jr, L., Shaik, S., Nam, W. Proc. Natl. Acad. Sci. USA
2007, 104, 19181
Sastri, V., Seo, M. S., Park, M. J., Kim, K. M., Nam, W. Chem. Commun. 2005, 1405
Schaefer, A., Huber, C., Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829
280
Scheidt, W. R., Cheng, B., Safo, M. K., Cukiernik, F., Marchon, J.-C., Debrunner, P. G.
J. Am. Chem. Soc. 1992, 114, 4420
Schindelin, H., Kisker, C., Schlessman, J.L., Howard, J.B., Rees, D.C. Nature, 1997,
387, 370
Schlichting, I., Berendzen, J., Chu, K., Stock, A. M., Maves, S. A., Benson, D. E., Sweet,
R. M., Ringe, D., Petsko, G. A., Sligar, S. G. Science 2000, 287, 1615
Schrock, R. R. Acc. Chem. Res. 1979, 12, 98
Schrödinger, E. Phys. Rev., 1926, 28, 1049
Schünemann, V., Gerdan, M., Trautwein, A. X., Haoudi, N., Mandon, D., Fischer, J.,
Weiss, R., Tabard, A., Guilard, R. Angew. Chem., Int. Ed. Engl. 1999, 38, 3181
Schwabe, T., Grimme, S. Acc. Chem. Res. 2008, 41, 569,
Schwabe, T., Grimme, S. Phys. Chem. Chem. Phys. 2007, 9, 3397
Shaik S., Chen H., Janardanan, D. Nature Chem. 2011, 3, 19
Shaik S., de Visser, S. P., Ogliaro, F., Schwarz H., Schröder, D. Curr. Opin Chem. Biol.
2002, 6, 556
Shaik S., Kumar D., de Visser, S. P. J. Am. Chem. Soc. 2008, 130, 10128
Shaik, S. S. J. Am. Chem. Soc. 1981, 103, 3692
Shaik, S., de Visser, S.P., Kumar, D. J. Bio. Inorg. Chem. 2004. 9, 6, 661
Shaik, S., de Visser, S. P., Kumar, D. J. Am. Chem. Soc. 2004, 126, 11746
Shaik, S., de Visser, S. P., Ogliaro, F., Schwarz, H., Schröder, D. Curr. Opin. Chem. Biol.
2002, 6, 556
281
Shaik, S., et al. Chem. Rev, 2005. 105,6, 2279
Shaik, S., Filatov, M., Schröder, D., Schwarz, H. Chem. Eur. J. 1983, 4, 193
Shaik, S., Hiberty, P. C. A Chemist’s Guide to Valence Bond Theory, Wiley-Interscience,
New York, NY, 2007
Shaik, S., Kumar, D., de Visser, S. P. J. Am. Chem. Soc. 2008, 130, 10128
Shaik, S., Kumar, D., de Visser, S. P., Altun, A., Thiel W. Chem. Rev. 2005, 105, 2279
Shaik, S., Lai, W., Chen, H., Wang, Y. Acc. Chem. Res. 2010, 43, 1154
Shanklin, J., Somerville, C. Proc. Natl. Acad. Sci. USA 1991, 88, 2510
Sharma, P. K., de Visser, S. P., Ogliaro, F., Shaik, S. J. Am. Chem. Soc. 2003, 125, 2291
Shelby, M.L., Mara M.W., Chen, L.X. Coord. Chem. Rev, 2014. 277–278, 291
Sheldrick, G. M. SADABS 2.0, 2000
Sheldrick, G. M. SHELXL-2014: Program for Crystal Structure Refinement, University
of Göttingen: Göttingen, Germany, 2014
Sherwood, P., et al. J. Mol. Struct. 2003, 632, 1
Shi, S., Klotz, U. Clinical Pharmacokinetics, 2012. 51, 2,77
Shinkyo, R., Xu, L., Tallman, K. A., Cheng, Q., Porter, N. A., Guengerich, T. P. J. Biol.
Chem. 2011, 286, 33021
Shokri, A., Abedin, A., Fattahi, A., Kass, S. R. J. Am. Chem. Soc. 2012, 134, 10646
Shook, R. L., Peterson, S. M., Greaves, J., Moore, C., Rheingold, A. L., A. Borovik, S. J.
Am. Chem. Soc. 2011, 133, 5810
282
Siegbahn, P. E. M., Blomberg, M. R. A. Chem. Rev. 2010, 110, 7040
Siegbahn, P. E. M., Blomberg, M. R. A. J. Phys. Chem. A 2008, 112, 12772
Siegbahn, P. E. M., Borowski, T. Acc. Chem. Res. 2006, 39, 729
Sil, D., Khan, F. S. T., Rath, S. P. Inorg. Chem. 2014, 53, 11925
Sisak, A., Ungváry, F., Kiss, G. J. Mol. Catal. 1983, 18, 223
Slater, C., The self-consistent field for molecular and solids. In Quantum Theory of
Molecular and Solids, McGraw-Hill, New York, 1974, Vol. 4.
Smith, W., Forester, T. R. J. Mol. Graphics 1996, 14, 136
Song, J., Seo, M. S., DeBeer George, S., Ohta, T., Song, R., Kang, M-J., Tosha, T.,
Kitagawa, T., Solomon, E. I., Nam, W. J. Am. Chem. Soc. 2007, 129, 1268
Song, W. J., Ryu, Y. O., Song, R., Nam, W. J. Biol. Inorg. Chem. 2005, 10, 294
Sono, M., Roach, M. P., Coulter, E. D., Dawson, J. H. Chem. Rev. 1996, 96, 2841
Stenkamp, R. E. Chem. Rev. 1994, 94, 715
Stephenson, N. A., Bell, A. T. Inorg. Chem. 2006, 45, 2758
Su, C.-J., Su, P.-C., Chi, Y., Peng, S.-M., Lee, G.-H. J. Am. Chem. Soc. 1996, 118, 3289
Su, T., Chesnavich, W. J. J. Chem. Phys. 1982, 76, 5183
Swart, M., Groenhof, A. R., Ehlers, A. W., Lammertsma, K. J. Phys. Chem. A 2004, 108,
5479
Takahashi, A., Kurahashi, T., Fujii, H. Inorg. Chem. 2011, 50, 6922
283
Takahashi, A., Yamaki, D., Ikemura, K., Kurahashi, T., Ogura, T., Hada, M., Fujii, H.
Inorg. Chem. 2012, 51, 7296
Takemoto, S., Morita, H., Karitani, K., Fujiwara, H., Matsuzaka, H. J. Am. Chem. Soc.
2009, 131, 18026
Tani, M. Matsu-ura, S. Nakayama, M. Ichimura, N. Nakamura, Y. Naruta, J. Am. Chem.
Soc. 2001, 123, 1133
Tao, J. M., Perdew, J. P., Staroverov, V. N., Scuseria, G. E. Phys. Rev. Lett. 2003, 91,
146401
Thomann, H., Bernardo, M., Goldfarb, D., Kroneck, P. M. H., Ullrich, V. J. Am. Chem.
Soc. 1995, 117, 8243
Thomas, J.M. Phy. Chem. Chem. Phys, 2014. 16, 17, 7647
Thorn D. L., Hoffmann, R. J. Am. Chem. Soc. 1978,100, 2079
Tomasi, J., Mennucci, B., Cammi, R. Chem. Rev., 2005, 105, 2999
Tomter, A. B., Zoppellaro, G., Andersen, N. H., Hersleth, H.-P., Hammerstad, M., Røhr,
Å. K., Sandvik, G. K., Strand, K. R., Nilsson, G. E., Bell III, C. B., Barra, A.-L., Blasco,
E., Le Pape, L., Solomon, E. I., Andersson, K. K. Coord. Chem. Rev. 2013, 257, 3
Tsipis, A.C. Coord. Chem. Rev, 2014. 272, 0, 1
TURBOMOLE V6.2 2010, a development of University of Karlsruhe and
Forschungszentrum Karlsruhe GmbH, 1989-2007, TURBOMOLE GmbH, since 2007,
available from http://www.turbomole.com.
Ueki, M., Matsumoto, Y., Jodry, J. J., Mikami, K. Synlett 2001, 1889
284
Ungváry, F. J. Metalorg. Chem. 1972, 36, 363
Usharani, D., Lai, W., Li, C., Chen, H., Danovich, D., Shaik, S. Chem. Soc. Rev. 2014,
43, 4968
Valdez, C. E., Smith, Q. A., Nechay, M. R., Alexandrova, A. N. Acc Chem Res 2014, 47,
3110
van Eldik, R. Coord. Chem. Rev. 2007, 251, 1649
van Leeuwen, P. W. N. M. Appl. Catal. A: General 2001, 212, 61
van Leeuwen, P. W. N. M., Morokuma, K., van Lenthe, J. H. Eds., Theoretical Aspects of
Homogeneous Catalysis, Kluwer Academic Publishers: Dordrecht, The Netherlands,
1995
van Leeuwen, P. W. N. M., Roobeek, C. F., Orpen, A. G. Organometallics, 1990, 9, 2179
van Leeuwen, P. W. N. M., Roobeek, C. F., van der Heijden, H. J. J. Am. Chem. Soc.
1994, 116, 12117
Van Leeuwen, P. W. N. M., Roobeek, C. F., Wife, R. L., Frijns, J. H. G. J. Chem. Soc.
Chem. Commun. 1986, 31
Vardhaman, A. K., Barman, P., Kumar, S., Sastri, C. V., Kumar, D., de Visser, S. P.
Angew. Chem. Int. Ed. 2013, 52, 12288
Vardhaman, A. K., Sastri, C. V., Kumar, D., de Visser, S. P. Chem. Commun. 2011, 47,
11044
Vardhaman, K., Barman, P., Kumar, S., Sastri, C. V., Kumar, D., de Visser, S. P. Angew.
Chem. Int. Ed. 2013, 52, 12288
285
Vardhaman, K., Sastri, C. V., Kumar, D., de Visser, S. P. Chem. Commun. 2011, 47,
11044
Vaska, L. Acc. Chem. Res. 1968, 1,335
Vaska, L., Diluzio, J. W. J. Am. Chem. Soc. 1962, 84, 679
Veitch, N. C., Smith, A. T. Adv. Inorg. Chem. 2000, 51, 107
Versluis, L., Ziegler, T., Fan, L. Inorg. Chem. 1990, 29, 4530
Vosko, H., Wilk L., Nusair M., Can. J. Phys., 1980, 58, 1200
Vougioukalakis, G. C., Grubbs, R. H. Chem. Rev. 2010, 110, 1746
Wachters, A. J. H. J. Chem. Phys. 1970, 52, 1033
Wang, L.-P., Wu, Q., van Voorhis, T. Inorg. Chem. 2010, 49, 4543
Watanabe, Y., Iyanagi, T., Oae, S. Tetrahedron Lett. 1980, 21, 3685
Watanabe, Y., Iyanagi, T., Oae, S. Tetrahedron Lett. 1982, 23, 533
Watanabe, Y., Nakajima, H., Ueno, T. Acc. Chem. Res. 2007, 40, 554
Watanabe, Y., Numata, T., Iyanagi, T., Oae, S. Bull. Chem. Soc. Jpn. 1981, 54, 1163
Weinberg, D. R., Gagliardi, C. J., Hull, J. F., Murphy, C. F., Kent, C. A., Westlake, B. C.,
Paul, A., Ess, D. H., McCafferty, D. G., Meyer, T. J. Chem. Rev. 2012, 112, 4016
Weinhold, F., Carpenter, J. E. The Natural Bond Orbital Lewis Structure Concept for
Molecules, Radicals, and Radical Ions. In The Structure of Small Molecules and Ions,
Naaman, R., Vager, Z., Eds., Plenum Press, New York, 1988, pp 227
286
Weiss, R., Fischer, J., Bulach, V., Schünemann, V., Gerdan, M., Trautwein, A. X.,
Shelnutt, J. A., Gros, C. P., Tabard, A., Guilard, R. Inorg. Chim. Acta. 2002, 337, 223
Weiss, R., Gold, A., Terner, J. Chem. Rev. 2006, 106, 2550
Wen, X., Wang, J.-S., Kivistö, K. T., Neuvonen, P. J., Backman, J. T. Br. J. Clin.
Pharmacol. 2001, 52, 547
Weymuth, T., Couzijn, E. P. A., Chen, P., Reiher, M. J. Chem. Theory Comput. 2014, 10,
3092
White, R. E., Groves, J. T. McClusky, G. A. Acta Biol. Med. Germ. 1979, 38, 475
Widenhoefer, R. A., Buchwald, S. L. J. Am. Chem. Soc. 1983, 120, 6504
Wilkins, P. C., Wilkins, R. G. Coord. Chem. Rev. 1987, 79, 195
Wong, S. D., Bell III, C. B., Liu, L. V., Kwak, Y., England, J., Alp, E. E., Zhao, J., Que
Jr, L., Solomon, E. I. Angew. Chem. Int. Ed. 2011, 50, 3215
Wu, A., Mayer, J. M. J. Am. Chem. Soc. 2008, 130, 14745
Yamada, M., Hagiwara, H., Torigoe, H., Matsumoto, N., Kojima, M., Dahan, F.,
Tuchagues, J.-P., Re, N., Iijima, S. Chem. Eur. J. 2006, 12, 4536
Ye, S., Neese, F. Proc. Natl. Acad. Sci. USA 2011, 108. 1228
Ye, S., Wu, X., Wei, L., Tang, D., Sun, P., Bartlam, M., Rao, Z. J. Biol. Chem. 2007, 282,
3391
Yeong, Y. J., Kang, Y., Han, A. R., Lee, Y. M., Kotani, H., Fukuzumi, S., Nam, W.
Angew. Chem., Int. Ed. 2008, 47, 7321
287
Yoon, J., Wilson, S. A., Jang, Y. K., Seo, M. S., Nehru, K., Hedman, B., Hodgson, K. O.,
Bill, E., Solomon, E. I., Nam, W. Angew. Chem. Int. Ed. 2009, 48, 1257
Yoshioka, S., Takahashi, S., Ishimori, K., Morishima, I. J. Inorg. Biochem. 2000, 81, 141
Yoshizawa, K., Shiota, Y., Yamabe, T. J. Am. Chem. Soc. 1983, 120, 564
Yu, F., Cangelosi, V. M., Zastrow, M. L., Tegoni, M., Plegaria, J. S., Tebo, A. G.,
Mocny, C. S., Ruckthong, L., Qayyum, H., Pecoraro, V. L. Chem. Rev. 2014, 114,3495
Yu, J., Hunt, F. Proc. Natl. Acad. Sci. USA 2009, 106, 14315
Yuan, X., Wang, Q., Horner, J. H., Sheng, X., Newcomb, M. Biochemistry 2009, 48,
9140
Zhao, Y., Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157
Zhao, Y., Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215