a computational study of free radicals in chemistry …
TRANSCRIPT
A COMPUTATIONAL STUDY OF FREE RADICALS IN
CHEMISTRY AND BIOLOGY
ADRIAN MATTHEW MAK WENG KIN (B.Sc.(Hons.)), NUS
A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
NUS GRADUATE SCHOOL OF INTEGRATIVE SCIENCES AND
ENGINEERING
NATIONAL UNIVERSITY OF SINGAPORE
2007
ii
Abstract
Free radicals are involved in many aspects of our daily lives, but due to their transient
nature and high reactivity, many free radicals and free radical precursors prove
difficult to study experimentally. The use of computational quantum chemical
methods to study these intriguing molecules has become popular today, given the
increasing availability and lowering cost of computing power. In this work,
computational studies on two families of reactive species are presented. First, a study
on intermediate species formed from the action of nitric oxide, NO•, with particular
emphasis on the direct effects of the free radical precursor peroxynitrite, ONOO–, and
the second, a study of bis(thiocarbonyl)disulfides (RCS2)2 in industrial sulfur
vulcanization of rubber, in particular, tetramethylthiuram disulfide (TMTD).
Knowledge of the intricate chemistry behind these reactive species will be useful in
designing experiments in this area, and better understanding of the way they work in
nature can thus be achieved.
iii
Acknowledgements The greatest thanks go to God, without whom all this would not have been possible.
Next, I am grateful to my family and Juliana, for being by my side as sound pillars of
support; for being there for me when I needed them most.
To Dr. Richard Wong, my thesis supervisor, thanks for the knowledge imparted to
this student of yours, and for being so much more than just a mere supervisor to me,
and also to my co-supervisor Dr. Matt Whiteman, for his valuable advice and help
along the way. Many thanks also go to Prof. Ralf Steudel and Dr. Yana Steudel of
Technical University of Berlin for opportunities in collaborative work.
Thanks goes to the Agency for Science, Technology and Research (A*STAR) for
providing the funding for graduate study, and the opportunity for a brighter future. To
those at NGS, here’s a big ‘thank you’ for your efficiency in administering to us
graduate students.
I am thankful for the many friends I made throughout the course of my study in the
Molecular Modeling Laboratory. To my friends, Henry, Jason, Ted, St. Francis Music
Ministry, Nicholas, Sabrina, Paul, Celene, Mr. & Mrs. Kodama, I am grateful to you,
for cheering me on.
Thank you, Mother Mary, for praying for me.
iv
Dedicated to my late grandfather.
v
List of Publications Included in this Thesis
The following papers were published as a direct consequence of the work undertaken
in this thesis:
1. Thermochemistry of Reactive Nitrogen Oxide Species and Reaction
Enthalpies for Decomposition of ONOO– and ONOOH. Mak, A. M.; Wong, M. W. Chem. Phys. Lett. 2005, 403, 192.
2. Homolytic Dissociation of the Vulcanization Accelerator Tetramethylthiuram Disulfide (TMTD) and Structures and Stabilities of the Related Radicals Me2NCSn
• (n = 1–4). Steudel, R.; Steudel, Y.; Mak, A. M.; Wong, M. W. J. Org. Chem. 2006, 71, 9302.
3. Reaction of the Radical Pair NO2• and CO3
•– with 2-[6-(4′-Amino)phenoxy-3H-xanthen-3-on-9-yl]benzoic Acid (APF). Mak, A. M.; Whiteman, M.; Wong, M. W. J. Phys. Chem. A, 2007, 111, 8202.
vi
Table of Contents Abstract ii
Acknowledgements iii
Table of Contents vi
Summary Xi
List of Tables 1
List of Figures 4
List of Reaction Schemes 7
1. Introduction 8
2. Theory and Methodology 16
2.1 Introduction 16
2.2 The Schrödinger Equation 16
2.3 Time-independence 17
2.4 The Hamiltonian Operator 18
2.5 Atomic Units 19
2.6 Born-Oppenheimer Approximation 19
2.7 Hartree-Fock Method 20
2.7.1 The Wave Function Φ as a Slater Determinant 20
2.7.2 Basis Sets 21
2.7.3 The Variation Principle 24
vii
2.7.4 The Fock Matrix 25
2.7.5 The Roothaan-Hall Equations 25
2.8 Restricted and Unrestricted Hartree-Fock Methods 27
2.9 Correlation Effects 29
2.9.1 Configuration Interaction (CI) 30
2.9.2 Limited Configuration Interaction 31
2.9.3 Coupled Cluster Methods 31
2.9.4 Møller-Plesset Perturbation Theory 32
2.10 Multi-Configuration SCF 34
2.11 Density Functional Theory 35
2.11.1 Local Density Approximation (LDA) and Local Spin Density
Approximation (LSDA) 36
2.11.2 Generalized Gradient Approximation (GGA) 37
2.11.3 Hybrid Functionals 39
2.12 Composite Methods for Accurate Thermochemistry 40
2.12.1 Gaussian-n Methods 40
2.12.2 CBS-n Methods 43
2.13 Methods of Calculation of ESR G-shifts 45
2.14 Solvation Models 46
2.14.1 Onsager Reaction Field Theory 47
2.14.2 Integral Equation Formalism for Polarizable Continuum
Solvent Model (IEFPCM) 48
3. Thermochemistry of Reactive Nitrogen Oxide Species 55
3.1 Introduction 55
viii
3.2 Computational Methods 57
3.3 Results and Discussion 59
3.3.1 Enthalpies of Formation of RNOS 59
3.3.2 Enthalpies of Reactions Involving ONOO– and ONOOH
Decomposition 66
3.4 Conclusion 68
4. Decomposition of Nitrosoperoxycarbonate Anion: Gas Phase and
Solvent Modeled Studies 73
4.1 Introduction 73
4.2 Computational Methods 75
4.3 Results and Discussion 77
4.3.1 Gas-phase Reaction Profile of ONOO– Reaction with CO2 77
4.3.2 Implicit Solvation Model (ISM) Results 82
4.3.3 Explicit Solvation Model (ESM) Results 84
4.3.4 Hybrid Solvation Model (HSM) Results 89
4.3.5 UV-Visible Spectra of ONOO–, ONOOCO2– and CO3
•– 91
4.4 Concluding Remarks 95
5. Reaction of the Radical Pair NO2• and CO3
•– with 2-[6-(4′-
Amino)phenoxy-3H-xanthen-3-on-9-yl]benzoic Acid (APF) 100
5.1 Introduction 100
5.2 Methods 103
5.2.1 Computational Methods 103
5.2.2 Experimental Methods 104
ix
5.3 Results and Discussion 106
5.3.1 Influence of HCO3− concentration on ONOO− mediated
oxidation of APF 106
5.3.2 Gas Phase Calculations 107
5.3.3 Solvent Effects 118
5.4 Concluding Remarks 118
6. Homolytic Dissociation of the Vulcanization Accelerator
Tetramethylthiuram Disulfide (TMTD) and Structures and Stabilities of
the Related Radicals Me2NCSn• (n = 1−4) 125
6.1 Introduction 125
6.2 Computational Methods 127
6.3 Results and Discussion 129
6.3.1 Structures and Stabilities of Tetramethylthiuram Sulfides and
Their Related Radicals 129
6.3.2 Impact of Metal Cations on the Dissociation of
Tetramethylthiuram Di- and Trisulfides 143
6.4 Concluding Remarks 153
7. Dissociation Enthalpies and Activation Barriers of Various
Bis(thiocarbonyl)disulfides R–C(=S)–S–S–C(=S)–R 159
7.1 Introduction 159
7.2 Computational Methods 162
7.3 Results and Discussion 163
7.3.1 Substituted bis(thiocarbonyl)disulfides, R = H, Me, F, Cl, OMe, 164
x
SMe, NMe2, PMe2
7.3.2 Novel bis(thiocarbonyl)disulfides, R = OSF5, Gu1, Gu2 174
7.4 Concluding Remarks 180
Appendix 1 183
xi
Summary
Free radicals are isolated atoms or molecules that possess one or more unpaired
electrons, and are capable of independent existence. Electrons within an atomic or
molecular system seek to be paired with another electron with opposite spin, resulting
in stable electronic configurations. Free radicals can react readily with many
molecules, in a myriad of mechanisms.
The study of free radicals is an interest of many research groups worldwide. This
comes as no surprise, since radicals play important roles in many chemical and
biological processes. For instance, nitric oxide (NO•) and superoxide (O2•–) play
important roles in biochemical pathways, triplet oxygen (O2) is a biradical that is
essential in combustion processes, benzoyloxyl radicals (PhCOO•) are used as
initiators in radical polymerization processes. Since most free radicals and free radical
precursors are transient by nature, the experimental study of them is challenging. The
enormous computing power available in the present day allows the use of
computational methods to study the chemistry of these intriguing compounds.
The work described in this thesis is based on computational studies of the molecular
properties, reactions and hence chemistry of free radicals and is centered on two
themes. The first is the study of reactive nitrogen species (RNS) and reactive oxygen
species (ROS), particularly those involving intriguing free radical precursor
peroxynitrite, ONOO–. The thermochemistry of reactions involving peroxynitrite was
evaluated and reviewed using high level ab initio and composite methods. As the
xii
most likely compound to react directly with ONOO– is CO2, a solvation study to
provide insight into solvent effects on the energetics and mechanisms of this reaction
was attempted, and its results presented. The detection and quantitation of RNS such
as ONOO– is carried out in a number of ways, of which, one method involves the use
of a pre-fluorescent probe in vivo in vitro. The mechanism of reaction of ONOO– with
such a fluorescent probe, APF, was studied, to provide a better picture of the reactive
species directly responsible for oxidation of the probe.
The second theme in this thesis is the study of bis(thiocarbonyl)disulfides (RCS2)2 and
thiocarbamate radicals RCS2• in the sulfur vulcanization of rubber. A theoretical study
on the energetics, molecular structures and properties of the intermediates involved in
tetramethylthiuram disulfide (TMTD) mediated sulfur vulcanization is presented.
Also presented is an exploration of novel bis(thiocarbonyl)disulfides to examine their
feasibility as alternatives to TMTD.
1
List of Tables PageTable 2.1. Components of CBS-4, CBS-q, CBS-Q, CBS-QCI/APNO, G3(MP2) and G3 models.
45
Table 3.1. Calculated standard enthalpies of formation (ΔH°f,298, kJ
mol–1) of sixteen RNOS using various Gaussian-n methods.
60
Table 3.2. Calculated standard enthalpies of formation (ΔH°f,298, kJ
mol–1) of sixteen RNOS using various CBS methods.
61
Table 3.3. Selected calculated bond lengths of RNOS in Ångstroms.
62
Table 3.4. Calculated structural parameters of planar cis-peroxynitrite, ONOO–, at various levels of theory.
64
Table 3.5. Calculated reaction enthalpies, in kJ mol–1, for reactions (1) to (3) at 298 K.
67
Table 4.1. Zero point corrected relative energies of 1-7, TS1 and TS2, in kJ mol–1.
82
Table 4.2. B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d) + ZPE and spin correction relative energies of 1-7, TS1 and TS2, in kJ mol–1, using implicit solvation models (ISM).
82
Table 4.3. B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d) + ZPE + Spin correction relative energies of 1a-7a, TS1a and TS2a, in kJ mol–1, using the explicit solvation model (ESM).
85
Table 4.4. B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d) + ZPE + Spin correction relative energies of 1a-7a, TS1a and TS2a, in kJ mol–1, using the hybrid solvation model (HSM).
89
Table 4.5. Predicted UV-vis absorption bands of cis-ONOO– (1), trans-ONOO– (2), cis-ONOOCO2
– (4), trans-ONOOCO2– (5), and
CO3•– (7) using various methods, based geometry obtained at
B3LYP/6-31+G(d) level.
91
Table 4.6. Calculated UV-Visible excitation wavelengths λexcitation, and corresponding oscillator strength values, of CO3
•– with 0, 1 and 2 explicit H2O molecules, and incorporating IEFPCM for modeling bulk effects of an aqueous solvent, using the TD-B3LYP method and the aug-cc-pVTZ basis set.
94
2
Table 5.1. Calculated activation barriers (kJ mol–1) of the two reaction pathways in the gas phase and in a polar medium at the MPW1K/6-311+G(d,p)//MPW1K/6-31G(d) level.
119
Table 6.1. Calculated Reaction Enthalpies (∆H°298, kJ mol–1) Using the G3X(MP2) and G3X(MP2)-RAD Methods.
135
Table 6.2. NBO Atomic Charges of the Donor Atoms and Dipole Moments μ in Debye of Species 1–4, 7, 7a and 7b.
139
Table 6.3. Binding energies, in kJ mol–1, and dipole moments μ, in Debye, of the lithium complexes 8–12, 14 and 15. Energies correspond to G3X(MP2) values, and dipole moments are calculated at the B3LYP/6-31G(2df,p) level.
146
Table 7.1. Selected calculated bond lengths (in Ångstroms) of (RCS2)2. Molecular geometry was optimized at the B3LYP/6-31G(2df,p) level of theory. Average experimental values for C—R bond lengths are included for comparison.
165
Table 7.2. Selected bond lengths (in Ångstroms) of RCS2•. Molecular
geometry was optimized at the B3LYP/6-31G(2df,p) level of theory.
167
Table 7.3. ∆H°298 and ∆G°298 for the homolytic S—S bond dissociation reaction (RCS2)2 → 2 RCS2
• calculated at the G3X(MP2) level of theory.
168
Table 7.4. NBO charges of (RCS2)2 and RCS2•, evaluated at the
B3LYP/6-31G(2df,p) level of theory, for substituents R = H, Me, F, Cl, OMe, SMe, NMe2 and PMe2.
169
Table 7.5. Selected calculated bond lengths (in Ångstroms) of [Li(RCS2)2]+. Molecular geometry was optimized at the B3LYP/6-31G(2df,p) level of theory.
170
Table 7.6. Selected calculated bond lengths (in Ångstroms) of [Li(RCS2)]•+. Molecular geometry was optimized at the B3LYP/6-31G(2df,p) level of theory.
170
Table 7.7. Selected calculated bond lengths (in Ångstroms) of triplet [(RCS2)Li(S2CR)]+. Molecular geometry was optimized at the B3LYP/6-31G(2df,p) level of theory.
171
Table 7.8. ∆H°298 and ∆G°298 for the coordination of (RCS2)2 and RCS2
• to Li+ in kJ mol–1, calculated at the G3X(MP2) level of theory.
172
Table 7.9. ∆H°298 and ∆G°298 for the reactions [Li(RCS2)2]+ → [Li(RCS2)]•+ + (RCS2)•, [Li(RCS2)2]+ → [(RCS2)Li(S2CR)]+, and [Li(RCS2)2]+ + Li+ → 2 [Li(RCS2)]•+ in kJ mol–1, calculated at the G3X(MP2) level of theory. 173
3
Table 7.10. Zero-point corrected reaction energies (∆E0) for the reaction (RCS2)2 → 2 RCS2
• , in kJ mol–1, for R = NMe2, OSF5, Gu1 and Gu2.
176
4
List of Figures PageFigure 2.1. A graphical representation of the variation of the energy of the 1s electron with its distance from the nucleus using a Slater type orbital (STO) basis function and a Gaussian type basis function (STO-3G basis set).
23
Figure 2.2. Flowchart summary of the procedure for solving the Roothaan-Hall equations.
27
Figure 2.3. Graphical representation of electron placement in orbitals following the RHF, UHF and ROHF formalisms.
29
Figure 2.4. Schematic diagram of electron correlation versus size of basis set.
34
Figure 2.5. A graph of eij(N) versus (N + δ)–1 for the helium PNOs.
44
Figure 3.1. Structures, element numbering, state and symmetry symbols of the sixteen RNOS: NO•, NO+, NO–, HNO, NO2
•, NO2+,
NO2–, HNO2, N2O, NO3
•, ONOO–, ONOOH, NOCl, NO2Cl, N2O3, N2O4.
57
Figure 4.1. Structures of local minima 1-7, TS1 and TS2, calculated at the B3LYP/6-31+G(d) level. Distances are given in Ångstroms.
79
Figure 4.2. Highest occupied molecular orbital (HOMO) of cis-ONOO– calculated at the B3LYP/6-31+G(d) level, showing extent of π*–π* interaction.
80
Figure 4.3. Structures of local minima 1a, 2a, 3a, 6a and 7a, calculated at the B3LYP/6-31+G(d) level. Distances are given in Ångstroms.
86
Figure 4.4. Structures of local minima 4a and 5a, and transition states TS1a and TS2a, calculated at the B3LYP/6-31+G(d) level. Distances are given in Ångstroms.
87
Figure 5.1. Plot of percentage yield of fluorescein, relative to ONOO–, versus the concentration of HCO3
– in the reaction mixture.
107
Figure 5.2. Calculated structural parameters of dianionic fluorescein (2E), neutral fluorescein, at the MPW1K/6-31G(d) level in the gas phase. X-ray crystal structural parameters are included for comparison.
110
5
Figure 5.3. Schematic gas-phase reaction energy profile for step 1 of pathway I, calculated at the MPW1K/6-311+G(d,p)//MPW1K/6-31G(d) +ZPE level with spin correction.
111
Figure 5.4. Schematic gas-phase reaction energy profile for step 2 of pathway I, calculated at the MPW1K/6-311+G(d,p)//MPW1K/6-31G(d) +ZPE level with spin correction.
113
Figure 5.5. Schematic gas-phase reaction energy profile for pathway II, calculated at the MPW1K/6-311+G(d,p)//MPW1K/6-31G(d) +ZPE level with spin correction.
114
Figure 5.6. MPW1K/6-31G(d) optimized geometries of 1A, 1B, 1D and 1E. Distances shown are in Ångstroms. Non essential hydrogens are omitted for clarity.
115
Figure 5.7. MPW1K/6-31G(d) optimized geometries of 2B, 2D, 3B and 3D. Distances shown are in Ångstroms. Non essential hydrogens are omitted for clarity.
116
Figure 5.8. MPW1K/6-31G(d) optimized geometries of transition structures TS1C, TS2A, TS2C, TS3A and TS3C. Distances shown are in Ångstroms. Non essential hydrogens are omitted for clarity.
117
Figure 5.9. Atom numbering scheme for 1A, 1E, TS2C and TS3C.
118
Figure 6.1. Chemical structure of tetramethylthiuram disulfide (TMTD).
125
Figure 6.2. B3LYP/6-31G(2df,p) optimized geometries of the two conformers of TMTD. Distances are in Ångstroms.
130
Figure 6.3. B3LYP/6-31G(2df,p) optimized geometries of the radicals Me2NCSn
• (n = 1–3). Distances are in Ångstroms.
131
Figure 6.4. UB3LYP/6-31G(2df,p) spin density plots of the Me2NCSn•
radicals (n = 1–3)
132
Figure 6.5. Chemical drawing of 2, showing resonance structures.
132
Figure 6.6. B3LYP/6-31G(2df,p) optimized geometries of the conformers of the Me2NCS4
• radical, 7, 7a and 7b, together with the numbering convention of sulfur atoms. Distances are given in Ångstroms.
139
Figure 6.7. UB3LYP/6-31G(2df,p) spin density plots of 7, 7a and 7b.
140
Figure 6.8. B3LYP/6-31G(2df,p) optimized geometries of TMTU, TMTM and TMTT. Distances are in Ångstroms.
142
6
Figure 6.9. B3LYP/6-31G(2df,p) optimized geometries of the cationic lithium complexes [Li(TMTD)]+ (8), [Li(SCNMe2)]•+ (9), [Li(S2CNMe2)]•+ (10) and [Li(S3CNMe2)]•+ (11). Distances are in Ångstroms.
144
Figure 6.10. B3LYP/6-31G(2df,p) optimized geometry of the triplet lithium complex [Li(S2CNMe2)2]+. Distances are given in Ångstroms. 148Figure 6.11. Schematic energy level diagram for the homolytic dissociation and complex formation reactions of TMTD with and without Li+ cations, based on G3X(MP2) computed reaction enthalpies.
149
Figure 6.12. B3LYP/6-31G(2df,p) optimized geometries of the triplet species 13 and 15 and singlet species 14. Distances are given in Ångstroms.
152
Figure 7.1. Examples of different sulfur vulcanization reagents commercially available.
160
Figure 7.2. The homolytic dissociation of (RCS2)2 with novel substituents R.
162
Figure 7.3. Typical structures of (RCS2)2, RCS2•, [Li(RCS2)2]+,
[Li(RCS2)]•+ and triplet [(RCS2)Li(S2CR)]+.
164
Figure 7.4. The B3LYP/6-31G(2df,p) optimized structure of (Me2PCS2)2, showing the lack of planarity at the C2PCS2 framework.
166
Figure 7.5. UB3LYP/6-31G(2df,p) spin density plots of the radicals RCS2
•.
167
Figure 7.6. B3LYP/6-31G(d) optimized structures of (F5SOCS2)2, ((Gu1)CS2)2 and ((Gu2)CS2)2. Distances are in Ångstroms.
175
Figure 7.7. B3LYP/6-31G(d) optimized structures of F5SOCS2•,
(Gu1)CS2• and (Gu2)CS2
•. Distances are in Ångstroms.
176
Figure 7.8. Schematic potential energy profile of the homolytic S—S bond dissociation reaction (RCS2)2 → 2 RCS2
•, for R = NMe2, OSF5, Gu1 and Gu2, calculated at the MP2/6-311+G(d,p)//B3LYP/6-31G(d) + ZPE level of theory. Relative energies are given in parentheses.
178
7
List of Reaction Schemes PageScheme 4.1. Proposed reaction scheme and labeling of various species involved in the reaction ONOO– + CO2 → ONO−OCO2
– → NO2• +
CO3•–.
77
Scheme 5.1. Action of RNS on APF.
102
Scheme 5.2. Pathway I of APF oxidation.
109
Scheme 5.3. Pathway II of APF oxidation.
109
8
Chapter One
Introduction
Free radicals are commonly defined as atomic or molecular species that possess one
or more unpaired electrons (i.e. nonzero spin), and are capable of independent
existence. These free radicals and other highly reactive closed-shell (i.e. all electrons
paired) molecules form a collective group known as reactive species (RS). The
considerable importance of RS in chemistry and biology has come to light only within
the last few decades, upon the realization that these RS are part of important chemical
reactions in the atmosphere, in industrial processes and in the human body. It is
certainly remarkable to see how free radical chemistry is involved in much of our
daily lives, for instance, the destruction of the ozone layer surrounding our planet, the
industrial preparation of plastics, and in cellular respiration and signaling [1].
Free radicals can be studied in a number of ways. Experimental methods employed to
study them include infra-red (IR) spectroscopy, mass spectrometry (MS), electron
paramagnetic resonance (EPR, also known as electron spin resonance, ESR), pulse
radiolysis and high speed electrochemistry [2]. Aside from these experimental
methods, ab initio or density functional theory calculations can be employed to study
the geometries, energies and hence molecular properties of these radicals in question.
The emergence of the field of computational chemistry followed the formulation and
development of quantum theory in the early twentieth century. Calculations based on
the fundamental principles that govern the behavior of chemical systems can be
9
carried out to determine their physical and chemical properties, thus revolutionizing
our understanding of physics and chemistry. Given the transient nature of most RS,
the use of computational tools to study their electronic structure has become a popular
method of scientific inquiry [3, 4]. We can now predict the structural and hence
molecular properties of these RS to a reasonably good degree of accuracy.
Chapter two of this thesis describes the theoretical framework behind the methods
employed in each topic, namely the ab initio methods, density functional methods,
and basis sets. The calculation of molecular properties such as thermochemistry and
electron spin resonance is briefly touched on in this chapter as well.
Following the description of the theoretical framework and methodology, the rest of
the work described in this thesis centers around two themes. First is the study of
reactive nitrogen and oxygen species that result from reactions of nitric oxide (NO•),
and the second, the study of bis(thiocarbonyl) disulfides ((RCS2)2) and related radicals
involved in the vulcanization of rubber by sulfur and sulfur-rich compounds.
Despite their reputation for causing damage, many biological processes are found to
be mediated by free radicals. The discovery of NO• as a biological mediator is
credited to the 1998 winners of the Nobel Prize for Physiology or Medicine:
Furchgott, Ignarro and Murad. Murad first observed in 1977 that NO• activates an
enzyme guanylyl cyclase and activates smooth muscle relaxation [5, 6]. Ignarro made
two key observations – that NO• has an effect of relaxing an artery [7] and this same
gas inhibits platelet aggregation and activates the cellular messenger cyclic guanosine
monophosphate (cGMP) [8]. Furchgott observed that the endothelium in smooth
10
muscles releases a factor, tentatively named endothelium-derived relaxing factor
(EDRF) that causes vasodilation [9]. It was established in 1986 that EDRF and NO•
are one and the same [10, 11]. This was a surprising discovery, given that during that
time, NO• was believed to be an atmospheric pollutant and little else.
The free radical NO• is formed by the action of nitric oxide synthases (NOS) on the
amino acid L-arginine [12], and can react with a host of other molecules to form
reactive nitrogen species (RNS). For instance, NO• can combine with superoxide
(O2•−), a free radical produced by phagocytic cells, to form peroxynitrite (ONOO–), an
essential precursor of other RNS such as NO2•. The direct and indirect effects of NO•
are diverse and can be cytoprotective, regulatory or cytotoxic [13-15]. Such diverse
effects of RNS sparked off tremendous research interest by many research groups
worldwide to better understand their biological behavior.
In chapter three, the thermochemistry of a series of biologically relevant reactive
nitrogen oxide species (RNOS) is discussed. The standard enthalpies of formation at
298K of NO•, NO+, NO–, HNO, NO2•, NO2
+, NO2–, HNO2, N2O, NO3
•, ONOO–,
ONOOH, NOCl, NO2Cl, N2O3, and N2O4 were computed using various composite
methods, namely the Gaussian-n (Gn) and complete basis set extrapolation (CBS)
methods, and the standard enthalpies of decomposition reactions involving ONOO–
and ONOOH were investigated at various levels of theory, namely the composite
methods mentioned above, and coupled-cluster theory. The goal of this work is to find
a suitable method that can well describe the thermochemistry of such nitrogen-
centered RS, and predict the standard heats of formation of ONOO– and ONOOH
which have not been experimentally ascertained.
11
A study on the effects of solvation on the reaction of cis- and trans-ONOO– with CO2
is described in chapter four. Formation of the nitrosoperoxycarbonate anion,
ONOOCO2–, and its subsequent decomposition into NO2
• and CO3•− free radicals have
been attributed as a source of free radicals in biological systems [16]. A
computational study of the above reaction using density functional theory was carried
out, to gain some chemical insight on the influence of an aqueous solvent on the
energetics involved. In addition, the transition energies of the RS ONOO–,
ONOOCO2–, and CO3
•− were investigated using a variety of theoretical methods, to
substantiate the assignments of their experimental UV absorption bands.
The detection and quantitation of RNS and reactive oxygen species (ROS) remains a
constant challenge to many scientists today. The use of fluorescent probes in various
assays is the most common method used today for such purposes. 2-[6-(4′-
Amino)phenoxy-3H-xanthen-3-on-9-yl]benzoic acid or 4-aminophenylfluorescein
(APF), developed by Setsukinai et al in 2003 [17], displayed excellent selectivity for
the RNS ONOO– over other RS such as superoxide (O2•−) and hydrogen peroxide
(H2O2). The mechanism of APF reaction with species related to ONOO– reaction with
CO2 was investigated, using a combination of experimental and computational
methods, to determine the identities of the species directly involved in the two-
electron oxidation of the probe, and the energetics involved. This work is described in
chapter five of this thesis.
Other than the involvement of free radicals in biological systems, free radicals have
also been employed in organic synthesis [18, 19] and other industrial applications.
12
Aside from being used for free radical chain substitution reactions in hydrocarbons
and free radical polymerization – common topics in most high school chemistry
textbooks – free radicals are also involved in functional group transformation (e.g.
decarboxylation, deamination, halogen reduction) [20], cyclization reactions and
aromatic substitutions. Reactive precursors such as bromine gas, tributyltin hydride
(Bu3SnH), and 2,2′-azobis(2-methylpropionitrile) (AIBN) are used to generate free
radicals in situ for such applications in organic synthesis. Free radicals can also be
used in cross-linking polymerization, of which one major industrial application would
be the sulfur vulcanization of rubber.
Sulfur vulcanization of rubber involves sulfide and disulfide cross-link generation
across neighboring poly(cis-isoprene) chains using elemental sulfur and accelerators
such as tetramethylthiuram disulfide (TMTD), thus providing strength to the overall
molecular structure. TMTD, having a formula (CH3)2NC(=S)SSC(=S)N(CH3)2 can be
thought of as a ‘reactive precursor’ as mentioned above, where the free radical
(CH3)2NCS2• is generated in situ. The cross-linking process is believed to take place
via various radical pathways. A detailed examination of the homolytic dissociation of
TMTD and the radicals involved in this process is described in chapter six. In addition,
the chemistry surrounding coordination of TMTD and associated radicals with a
lithium cation is explored. The thermochemistry surrounding their reactions are
derived using high-level composite methods, and calculation of molecular properties
such as spin density and electron spin resonance spectra are also covered in this study.
The work done in this chapter is a collaborative effort with Professor Ralf Steudel and
Dr. Yana Steudel from the Institute of Chemistry in the Technical University of Berlin.
13
While TMTD is already an adequate choice for an accelerator in rubber vulcanization,
there is still reliance on zinc oxide to facilitate the formation of thiocarbamoyl
radicals RCS2• as the homolytic S—S bond dissociation enthalpy in room temperature
conditions is about 150 kJ mol–1. In chapter seven, an exploration of other substituted
bis(thiocarbonyl) disulfides (RCS2)2 is attempted, to find suitable alternative rubber
vulcanization accelerators. The homolytic S—S bond dissociation enthalpy of
(RCS2)2, and coordination enthalpy of Li+ ions to (RCS2)2 and RCS2• for R = H, Me, F,
Cl, OMe, SMe, NMe2 and PMe2 were studied and compared. Additionally, the S—S
bond dissociation enthalpy and activation barriers for a number of novel substituents
R were examined, in an effort to find suitable substituents with can stabilize the RCS2•
radical and result in a more energetically favorable generation of these radicals.
References
1. Parsons, A. F., An Introduction to Free Radical Chemistry. Blackwell Science:
Oxford, 2000.
2. Fossey, J.; Lefort, D.; Sorba, J., Free Radicals in Organic Chemistry. Masson:
Paris, 1995.
3. Anders, M. W.; Yin, H.; Jones, J. P., Application of Computational Chemistry
in the Study of Biologically Reactive Intermediates In Advances in
Experimental Medicine and Biology: Biological Reactive Intermediates V:
Basic Mechanistic Research in Toxicology and Human Risk Assessment,
Snyder, R.; Kocsis, J. J.; Jollow, D. J.; Witmer, C. M.; Greim, H.; Sipes, I. G.;
14
Monks, T. J.; Kalf, G. F., Eds. Plenum Press: New York, 1996; Vol. 387, pp
347.
4. Ventura, O. N.; Kieninger, M. Pure Appl. Chem. 1998, 70, 2301.
5. Arnold, W. P.; Mittal, C. K.; Katsuki, S.; Murad, F. Proc. Natl. Acad. Sci. U. S.
A. 1977, 74, 3203.
6. Katsuki, S.; Arnold, W.; Mittal, C.; Murad, F. J. Cyclic Nucleotide Res. 1977,
3, 23.
7. Gruetter, C. A.; Barry, B. K.; McNamara, D. B.; Gruetter, D. Y.; Kadowitz, P.
J.; Ignarro, L. J. J. Cyclic Nucleotide Res. 1979, 5, 211.
8. Ignarro, L. J.; Lippton, H.; Edwards, J. C.; Baricos, W. H.; Hyman, A. L.;
Kadowitz, P. J.; Gruetter, C. A. J. Pharmacol. Exp. Ther. 1981, 218, 739.
9. Furchgott, R. F.; Zawadzki, J. V. Nature (London, U. K.) 1980, 288, 373.
10. Ignarro, L. J.; Buga, G. M.; Wood, K. S.; Byrns, R. E.; Chaudhuri, G. Proc.
Natl. Acad. Sci. U. S. A. 1987, 84, 9265.
11. Palmer, R. M. J.; Ferrige, A. G.; Moncada, S. Nature (London, U. K.) 1987,
327, 524.
12. Palmer, R. M. J.; Ashton, D. S.; Moncada, S. Nature (London, U. K.) 1988,
333, 664.
13. Espey, M. G.; Miranda, K. M.; Thomas, D. D.; Xavier, S.; Citrin, D.; Vitek, M.
P.; Wink, D. A. Ann. N. Y. Acad. Sci. 2002, 962, 195.
14. Miranda, K. M.; Espey, M. G.; Jourd'heuil, D.; Grisham, M. B.; Fukuto, J. M.;
Feelisch, M.; Wink, D. A., The Chemical Biology of Nitric Oxide. In Nitric
Oxide : Biology and Pathobiology, Ignarro, L. J., Ed. Academic Press: San
Diego, 2000; pp 41.
15. Wink, D. A.; Mitchell, J. B. Free Radical Biol. Med. 1998, 25, 434.
15
16. Veselá, A.; Wilhelm, J. Physiol. Res. 2002, 51, 335.
17. Setsukinai, K.-i.; Urano, Y.; Kakinuma, K.; Majima, H. J.; Nagano, T. J. Biol.
Chem. 2003, 278, 3170.
18. Barton, D. H. R. Tetrahedron 1992, 48, 2529.
19. Jasperse, C. P.; Curran, D. P.; Fevig, T. L. Chem. Rev. (Washington, DC, U. S.)
1991, 91, 1237.
20. Ramaiah, M. Tetrahedron 1987, 43, 3541.
16
Chapter 2
Theoretical Background and Methodology
2.1 Introduction
This chapter outlines the theoretical framework used in the quantum chemical
calculations that have been carried out. Sections 2.2 to 2.10 describe the theoretical
framework of ab initio methods and basis sets. Section 2.11 describes density
functional theory (DFT) and framework. Sections 2.12 to 2.13 describe methods used
to calculate and hence predict molecular properties such as thermochemistry, and
electron spin resonance (ESR) g-shifts. Section 2.14 briefly describes the methods
used to model solvation. Details of the intermediate steps in deriving equations can be
found in most quantum chemistry textbooks [1–3], and are included in Appendix 1.
2.2 The Schrödinger Equation
The time dependence of any quantum chemical system is described by the time-
dependent Schrödinger equation [4, 5]:
),(),(ˆ
00 txt
itxH Ψ∂∂
=Ψ (2-1)
17
Where ħ is the Planck’s constant divided by 2π, m is the mass of the system, V is a
potential energy operator and Ψ0( x ,t) is the wave function of the system as a function
of position and spin, x and time t.
2.3 Time-independence
In a real world problem, we are interested in the variation of the energy of the states
of a molecular system with x .
Assuming that the wave function Ψ( x ,t) can be factorized into two functions, one
being a function of x and one being a function of t, the factorized equations can be
substituted into equation (2-1) to yield the time independent Schrödinger equation.
The detailed steps can be found in section A1.1 of Appendix 1.
)()(ˆ xExH Ψ⋅=Ψ (2-2)
Ĥ is the time-independent Hamiltonian operator.
Solving the time-independent Schrödinger equation yields E, the energy term of the
state of interest, and the state wave function as a function of x . A knowledge of the
ground state wave function, Ψ0( x ), provides information on the physical properties of
the system.
18
An exact analytical solution of the Schrödinger equation for any system larger than
that of the hydrogen atom is too complicated, and currently not possible. However, a
number of assumptions and approximations can be employed to yield an approximate
solution to the Schrödinger equation.
2.4 The Hamiltonian Operator
The molecular Hamiltonian operator is made up of a kinetic energy and a potential
energy term, T and V respectively:
VTH ˆˆˆ += (2-3)
where
∇∑−= 2
2
2ˆ
kk km
T (2-4)
jk
kj
j jk ree
V ⋅= ∑∑< 04
1ˆπε
(2-5)
An atom or molecule is considered as a system of interacting nuclei and electrons.
The Hamiltonian operator for such a system is constructed as follows:
AB
BAM
A
M
ABij
N
i
N
ijiA
AN
i
M
AA
M
A Ai
N
i e ReZZ
re
reZ
MmH
i
2
1 0
21
1 1 0
2
1 1 0
2
1
22
1
2
41
41
41
22ˆ ∑∑∑∑∑∑∇∑∇∑
= >
−
= +== ===
++−−−=πεπεπε
Hkin
elecˆ= H
kin
nucˆ+ H
pot
nucelecˆ
−+ H
pot
elecelecˆ
−+ H
pot
nucnucˆ
−+
(2-6)
19
The above terms in the Hamiltonian correspond to the operators for kinetic energy
contribution of the electrons, kinetic energy contribution of the nuclei, potential
energy contribution from electron-nuclei interaction, potential energy contribution
from electron-electron interaction, and potential energy contribution from nuclei-
nuclei interaction respectively.
2.5 Atomic Units
While dealing with atomic and molecular systems, it is often useful to use atomic
units to simplify the results of our calculations. By setting , me, |e| and 4πε0 to equal
to 1 [6, 7], we obtain a less complicated expression for Ĥ:
∑∑∑∑∑∑∑∇∑∇= >
−
= +== ===
++−−−=M
A
M
AB AB
BAN
i
N
ij ij
N
i
M
A iA
AM
AA
N
ii R
ZZrr
ZH1
1
1 11 11
221
1
221 1ˆ
(2-7)
Energies are then measured in Hartrees, where 1 Hartree is defined to be the
Coulombic repulsion between two electrons separated by a distance of 1 Bohr radius,
or 0.52911725 Å. One Hartree is equivalent to 27.2107 eV or 4.35974394 × 10–18 J.
2.6 Born-Oppenheimer Approximation
The physical idea behind the Born-Oppenheimer approximation is that electrons move
much faster than nuclei, since the mass of a nucleus is typically thousands of times
that of an electron. For a system of nuclei with initial fixed positions, the electrons
20
adjust their positions with respect to these nuclei before the nuclei actually move [8].
By separation of nuclear and electronic motions, the problem can be simplified to the
electronic Schrödinger equation.
eleceffelecelec EH Ψ=Ψˆ (2-8)
The detailed steps can be found in section A1.2 of Appendix 1.
2.7 Hartree-Fock Method
Despite the approximations mentioned so far, the electronic Schrödinger equation (2-
8) for a molecule is a complex differential equation, and still cannot be solved
analytically except for the smallest systems.
The Hartree-Fock (HF) method [9, 10], or the self consistent field (SCF) method is a
numerical iterative technique employed to calculate the Slater determinant and an
expectation value of the electronic Hamiltonian. It uses a number of approximations
and assumptions to provide an approximate solution to the electronic Schrödinger
equation. The following subsections outline the steps taken in the method.
2.7.1 The wave function Φ as a Slater determinant
The electronic wave function Ψ should have two important properties. First, it should
21
be normalized, and second, because electrons are fermions, it should also be
antisymmetric, that is it should change sign when two electrons are interchanged.
An approximate electronic closed shell wave function, the Slater determinant, is
worked out in section A1.3 of Appendix 1. The Slater determinant yielded fulfills the
above requirements of being normalized and antisymmetric.
≈Ψ )(xe )(xeΦ
= )()()()()()()()()()()()(
)2()()2()()2()()2()()2()()2()()1()()1()()1()()1()()1()()1()(
!1
222211
222222222121
121212121111
NxNxNxNxNxNx
xxxxxxxxxxxx
NNNNNNNNN
NN
NN
βϕαϕβϕαϕβϕαϕ
βϕαϕβϕαϕβϕαϕβϕαϕβϕαϕβϕαϕ
(2-10)
The essential assumption made here in the Hartree-Fock method is that the wave
function of an electron, which is not a physical observable, can be mathematically
represented as an antisymmetric product of molecular orbitals. Each electron here is
assumed to experience an average charge distribution due to other electrons in the
system, and not interact explicitly with each and every one of the other electrons.
2.7.2 Basis Sets
Individual molecular orbitals φ in the Slater determinant (eq. 2-10) are expressed as
linear sums of normalized basis functions, χ. Basis functions are simply atomic
orbitals described by a mathematical function based on spherical or Cartesian
coordinates.
22
∑=
=Nbasis
ii c1μ
μμ χϕ (2-11)
Slater-type orbitals (STOs) were historically used as basis functions of molecular
orbitals because of their similarity to the atomic orbitals of the hydrogen atom.
),(),,;,,,( 1 φθφθζχ ζlm
rni YeNrrmln −−= (2-12)
N is a normalization constant; r, θ , φ are spherical coordinates; n, l, m are principal,
angular momentum and magnetic quantum numbers respectively; ζ is called an
“exponent”, and Ylm is a function describing the shape (the angular momentum part).
STOs have the advantage of describing wavefunctions well. The exponential term, e-ζr,
well describes the energy of electrons near the nucleus. However they also have the
disadvantage of being unsuitable for calculating two-electron integrals quickly. A
solution to this is the implementation of Gaussian-type functions (GTFs).
nmlzyx zyxNezyxnmlg )( 222
),,;,,,( ++−= αα
nmlr zyxNe2α−= (2-13)
N is a normalization constant; x, y, and z are spherical coordinates; n, l, m are integral
exponents at Cartesian coordinates, and α is called the “exponent” here.
GTFs allow fast calculation of two electron integrals, but representing an atomic
orbital with a GTF leads to error as GTFs do not have a cusp at the nucleus like STOs
do. However, addition of a number of GTFs yields a good approximation to an STO,
and is still computationally cheaper than the integration of the STO itself. GTFs are
23
grouped, or contracted, in linear combinations to form basis functions. A comparison
of an STO with an STO-3G (STO approximated by 3 GTFs) basis set for the 1s
orbital of the hydrogen atom is graphically presented in Figure 2.1.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 0.5 1 1.5 2
Distance, Angstroms
Figure 2.1. A graphical representation of the variation of the energy of the 1s electron with its distance from the nucleus using a Slater type orbital (STO) basis function and a Gaussian type basis function (STO-3G basis set). The filled line represents the trend for the STO, while the dotted line represents the trend for the STO-3G basis.
Basis sets can be grouped into a number of categories – minimal, split valence,
polarized basis sets, and basis sets with diffuse functions.
Minimal basis sets such as STO-3G use fixed-size atomic-type orbitals. They contain
the minimum number of basis functions needed for each atom.
Split valence basis sets have a larger number of basis functions per atom. Two or
more sizes of basis functions are assigned to the valence shells of each atom, allowing
the orbitals to change size. A double-split valence basis set, such as 3-21G, has the
24
valence shells of each atom split into two basis functions, the first basis function is a
contraction of two Gaussian primitives, and the second basis function consists of one
Gaussian primitive. Within the split valence basis set category, there are also n-ζ type,
where molecular orbitals are formed from linear combinations of n sizes of functions
for each atomic orbital, for example the double-ζ Dunning-Huzinaga (D95), and the
correlation consistent, polarized valence triple-ζ (cc-pVTZ) basis sets.
Polarized basis sets involve adding orbitals of a higher angular momentum than the
ground state electronic configuration to the description of each atom. This allows
orbitals to effectively change shape. Examples of polarized basis sets are 6-31G(d)
and 6-31G(d,p).
Diffuse functions can be added to basis sets to allow orbitals to occupy larger regions
in space. This allows better description of species where electrons are further away
from the nucleus, such as anions, and molecules with lone pairs of electrons.
Examples of such basis sets are 6-31+G(d) and 6-311++G(d,p).
2.7.3 The Variation Principle
The variation principle is a starting point for almost all methods that seek to find an
approximate solution to the Schrödinger equation. It states that for any acceptable
normalized function Φ, the expectation value of Ĥ will be an upper bound to the
lowest eigenvalue of Ĥ.
25
A consequence of the variation principle is that as the trial wave function, Φtrial,
approaches the true wave function Ψ0, the expectation value of Ĥelec, Etrial, approaches
the true energy of the system, E0.
2.7.4 The Fock Matrix
The Fock matrix, F, is an approximation of the true Hamiltonian operator of the
system. It is made up of Fock operators, )(ˆ iF , with the following general formula
(see section A1.4 of Appendix 1 for derivation)
∑=
−+=N
jijj
corei KJHiF
1)( )]}ˆˆ2{ˆ)(ˆ
(2-14)
coreiH )(
ˆ is the core Hamiltonian operator, jJ and ijK are Coulomb and exchange
operators respectively.
The equations that connect the electronic Hamiltonian, eH , and the orbital wave
functions φi can now be written in terms of the Fock operator F , φi, and a set of
constants, εi. in a form that resembles the Schrödinger equation. These equations are
also commonly referred to as the Hartree-Fock equations.
iiiF ϕεϕ =ˆ (2-15)
2.7.5 The Roothaan-Hall equations
Roothaan and Hall independently devised a method to represent the Hartree-Fock
26
equations using non-orthonormal Gaussian type or Slater type basis sets [11, 12]. The
Roothaan-Hall equations are:
∑=
=−Nbasis
ii CSF1
0)(ν
νμνμν ε ;
for μ = 1,2,…,Nbasis.
(2-16)
The electronic energy can be calculated from the elements of the core Hamiltonian
matrix coreH μν , the Fock matrix Fμν and the one electron density matrix Pμν. The nuclear
repulsion energy is then added to obtain the HF energy.
The Hartree-Fock energy is obtained by adding Eelec to the nuclear repulsion energy.
∑∑∑ ∑= <= =
++=M
A
M
BA AB
BANbasis Nbasis
coreHF R
ZZHFPE
11 1)(
21
μ νμνμνμν
(2-17)
The program structure for solving the Roothaan-Hall equations [1] is summarized in
Figure 2.2. The detailed derivation of the Roothaan-Hall equations is provided in
Section A1.5 of Appendix 1.
27
Form Fock Matrix
Solve Roothaan-Hall equations
∑=
=−Nbasis
ii CSF1
0)(ν
νμνμν ε
Form density matrix
∑=
=occ
iii CCP
1
*2 σλλσ
Calculate energy
)(21
1 1∑ ∑= =
+=Nbasis Nbasis
coreelec HFPE
μ νμνμνμν
Initial guess Pμν
Eelec and/or Pμνconverged?
Pμν
No
)]|(21)|[(
1 1
νσμλλσμνλ σ
λσμνμν ∑ ∑= =
−+=Nbasis Nbasis
core PHF
Calculate 2-electronIntegrals
Calculate 1-electronIntegrals
Sμν
coreH μν
(μν|λσ)
Done
Form Fock Matrix
Solve Roothaan-Hall equations
∑=
=−Nbasis
ii CSF1
0)(ν
νμνμν ε
Form density matrix
∑=
=occ
iii CCP
1
*2 σλλσ
Calculate energy
)(21
1 1∑ ∑= =
+=Nbasis Nbasis
coreelec HFPE
μ νμνμνμν
Initial guess Pμν
Eelec and/or Pμνconverged?
Pμν
No
)]|(21)|[(
1 1
νσμλλσμνλ σ
λσμνμν ∑ ∑= =
−+=Nbasis Nbasis
core PHF
Calculate 2-electronIntegrals
Calculate 1-electronIntegrals
Sμν
coreH μν
(μν|λσ)
Done
Figure 2.2. Flowchart summary of the procedure for solving the Roothaan-Hall equations.
2.8 Restricted and Unrestricted Hartree-Fock methods
The Restricted Hartree Fock (RHF) method described in the previous section assumes
a closed-shell system where each spatial orbital is doubly occupied. For open-shell
systems where there are singly occupied orbitals, the RHF method is not suitable.
28
The common method used to treat open-shell systems is to separate α and β electrons
into singly occupied spatial orbitals, and solve for two sets of Roothaan-Hall
equations, known as the Pople-Nesbet equations [13]:
FαCα = SCαεα (2-18)
FβCβ = SCβεβ (2-19)
This approach is termed as the spin-unrestricted Hartree-Fock (UHF) method.
The major drawback of the UHF method is that one-electron Slater determinants for α
and β electrons are not satisfactory eigenfunctions of the total spin operator, Ŝ2. Thus
the total wave function may be contaminated by wave functions of higher states; for
example, a doublet state contaminated with quartet or higher spin states. The spin
contamination – the deviation of the expectation value of the total spin operator, ⟩⟨ 2S ,
from S(S+1) where S is the 0.5 times the number of unpaired electrons – gives an
indication of the contamination from higher spin states.
Another modification to the RHF method to accommodate open-shell systems is the
Restricted Open-Shell Hartree-Fock (ROHF) method. The ROHF method follows the
RHF method as much as possible, and does not suffer from spin contamination from
higher spin states. However, the Fock matrix for the ROHF method is not unique, and
different Fock matrices can yield the same wave function.
A graphical representation of the orbital treatment for RHF, UHF and ROHF methods
is given in Figure 2.3.
29
RHF UHF ROHF
Figure 2.3. Graphical representation of electron placement in orbitals following the RHF, UHF and ROHF formalisms.
2.9 Correlation Effects
Throughout the formulation of the HF method, one assumption held was that each
electron moves in a time-averaged potential field caused by the other electrons in the
system. In reality, the motions of electrons are correlated, since they actually repel
each other individually, rather than each one of them reacting to an average field
generated by the other electrons. The energy calculated using the HF method is higher
than that when electron correlation is explicitly considered in the wave function. This
energy difference is known as the correlation energy.
ncorrelatioHFexact EEE += (2-20)
There are many techniques that are used to calculate this correlation energy, and are
collectively termed post SCF methods. The following subsections describe their
formulation.
30
2.9.1 Configuration Interaction (CI)
In the HF method, the wave function was a single determinant for a closed shell
system. To implement configuration interaction, one has to take into account other
electronic configurations that can be adopted by the same number of electrons in the
system. Taking the HF single determinant wave function Φ0 as reference, the
eigenvector corresponding to the exact solution of the Schrödinger equation, ΨCI, for a
given one-electron basis set, is written as a linear combination of determinant wave
functions (or configuration state functions, CSFs) from all electronic states, Φ:
…+Φ+Φ+Φ=Ψ ∑∑∑∑
< <
occ
ji
virt
qp
pqij
pqij
occ
i
virt
p
pi
piCI aaa 00
(2-21)
The above expression is the eigenvector for the full CI (FCI) method, wherein all
possible excited configurations have been included. The single-indexed term
∑∑ Φocc
i
virt
p
pi
pia represents the contributions from single excitations, and the double-
indexed term ∑∑< <
Φocc
ji
virt
qp
pqij
pqija represents the contributions from double excitations. The
use of the infinite expansion ΨCI in ⟩ΨΨ⟨ CIeCI H |ˆ| would yield the sum of HF
energy and all energy due to explicit electron correlation within the given basis set.
The disadvantage of carrying out a full CI calculation is that it will involve a great
number of CSFs to be computed, and is not practical except for very small systems.
31
2.9.2 Limited Configuration Interaction
Limited CI is carried out by adding only a limited set of substitutions to the HF
determinantal wave function. The CIS, CID and CISD methods (see section A1.6 of
Appendix 1) involve adding single, double, and both single and double substitutions
to the CI wave function respectively. However, these methods suffer the disadvantage
of not being size-consistent.
The quadratic configuration interaction method, using singles and doubles substitution
(QCISD) was developed by Pople et al [14] to ensure size consistency in the resulting
total energy. The QCISD equations are described in section A1.7 in Appendix 1.
The full triples substitution (QCISDT method) is also possible, but is computationally
impractical for large systems. To address this, a partial triples correction based on
perturbation at the singles-doubles level, ΔET is included in the QCISD formulation to
yield improved QCISD(T) energies [14].
2.9.3 Coupled Cluster Methods
Coupled cluster methods, derived by Purvis III et al [15], work on the ansatz that the
wave function can be expressed as exp( T )Ψ0, where T is the cluster operator. The
formulation of the coupled cluster method using singles and doubles substitution
(CCSD) equations is included in section A1.8 of Appendix 1.
32
There are a number of different formulations to address the triples correction for the
CCSD method [16-18]. One commonly used is that proposed by Raghavachari et al,
which is similar to that adopted in the QCISD(T) method, to yield CCSD(T) energies
[18].
2.9.4 Møller-Plesset Perturbation Theory
An alternative way to treat electron correlation is using Møller-Plesset perturbation
theory (MPPT) [19], whose principles are drawn from techniques in many body
perturbation theory (MBPT).
In MPPT, the Hamiltonian operator is first divided into two parts:
VHH ˆˆˆ0 λ+= (2-22)
Where Ĥ0 is the sum of one-electron Fock operators, which is soluble exactly, and
Vλ is a small perturbation. According to Rayleigh-Schrödinger perturbation theory
(RSPT), the ground state wave function and energy of the system can be expanded as
a power series:
…)3(3)2(2)1()0( Ψ+Ψ+Ψ+Ψ=Ψ λλλ (2-23)
…)3(3)2(2)1()0( EEEEE λλλ +++= (2-24)
Substituting these equations back into the time-independent Schrödinger equation
gives:
))(())(ˆˆ( )()()(
0 ∑∑∑ Ψ=Ψ+n
i
iin
i
iin
i
ii EVH λλλλ (2-25)
33
For n=0, the sum of the electronic orbital energies εi is obtained and for n=1, the HF
energy EHF = E(0) + E(1) and the HF wave function Ψ(0) are obtained.
Solving these equations to the nth order will yield the corresponding MPn energies.
The detailed derivation of the expressions of the MPn energies is provided in section
A1.9 of Appendix 1. MP2 energies include contributions where doubly excited
configurations interact with the ground state configuration. MP3 energies include
contributions of doubly excited configurations interacting with each other. MP4
energies include energy contributions of interactions that involve single, double and
quadruple excitations.
It is noted that MPPT is size consistent, but not variationally bound. It cannot be
assumed that adopting higher orders of perturbation will yield energies that are
progressively closer to the true energy.
Figure 2.4 summarizes the effects of correlation versus the effects of basis sets in
attempts to solve the Schrödinger equation using ab initio techniques. The horizontal
axis depicts the extent of electron correlation considered, and the vertical axis depicts
the size of the basis set. The true solution of the Schrödinger equation is reached when
a full configuration interaction using a complete basis set is performed.
34
Increasing electron correlation HF MP2 MP3 MP4 QCISD QCISD(T) … Full CI
Minimal e.g. STO-3G
Split-valence e.g. 3-21G
Polarized e.g. 6-31G(d)
Diffuse e.g. 6-311+G(d)
High angular momentum e.g. 6-311+G(2df,p)
…
Increasing size of basis set Complete basis set HF
limit Exact
Solution Ψ=Ψ EH
Figure 2.4. Schematic diagram of electron correlation versus size of basis set.
2.10 Multi-Configuration SCF
Instead of considering just a single electronic configuration in the construction of the
HF wave function from the Slater determinant, the multi-configuration SCF (MCSCF)
method considers multiple electronic configurations in the construction of the
electronic wave function [20]. Essentially it is an extension of the single configuration
HF method. The most common approach is the complete active space SCF (CASSCF)
method [21-23], in which a set of chemically important molecular orbitals from the
HF wave function – the active space – are chosen, and a full CI carried out within this
active space.
35
2.11 Density Functional Theory
Density Functional Theory (DFT) methods are in principle an exact theory of
electronic structure, with its base on the electron density ρ(r) instead of the
approximate many electron wave function Ψ[24]. DFT methods are extremely popular
today due to its versatility and reduced computational costs compared to conventional
ab initio methods. A great deal of information on DFT is available in most textbooks
[25, 26]; only a brief overview of DFT will be provided in this section.
Modern DFT methods stem from the Hohenberg-Kohn theorem [27], which states that
all properties of a system defined by an external potential, Vext, are uniquely
determined by the ground state electron density )(rρ . Hence, as a result of the
variational principle, the )(rρ that gives the minimum total energy is the ground state
electron density.
The Hamiltonian of a given system, and its electronic energy, is partitioned using the
Kohn-Sham approach [28].
H eeNe UTV ˆˆˆ ++= (2-26)
)]([ rE ρ
)]([)(
)]([)]([)]([
rFrdVr
rErTrE
Ne
eeNe
ρρ
ρρρ
+=
++=
∫
(2-27)
)]([ rF ρ is a functional of the electron density )(rρ , as Ψ is. The components of
)]([ rF ρ are: )]([ rTs ρ , the kinetic energy of the non-interacting system with density
36
)(rρ , )]([ rJ ρ , the classical electrostatic electron-electron repulsion energy, and the
remainder, )]([ rEXC ρ , the exchange-correlation energy.
The entire expression for the energy calculated using the DFT approach is written in
equation (2-28). A detailed explanation is provided in section A1.10 of Appendix 1.
=)]([ rE ρ ∑ ⟩∇⟨−
N
iii ϕϕ ||
21 2
)]([
|)(|
|)(|1|)(|21
12
11
212
212
21
rE
rdrrZ
rdrdrr
r
XC
N
i
M
Ai
A
A
j
N
i
N
ji
ρ
ϕ
ϕϕ
+
−
+
∑∫∑
∑∑∫∫
(2-28)
The exact form of )]([ rEXC ρ is unknown. Finding accurate forms of )]([ rEXC ρ is
still an agendum of modern research in quantum chemical methods.
2.11.1 Local Density Approximation (LDA) and Local Spin Density Approximation
(LSDA)
The center of the LDA and LSDA is the idea of a hypothetical uniform electron gas,
with number of electrons N and volume of the gas V. As N and V approach infinity,
the electron density ρ attains a constant value everywhere. EXC can then be assumed to
take the following form:
rdrrE XCLDAXC ))(()(][ ρερρ ∫= (2-29)
37
))(( rXC ρε is the exchange-correlation energy per particle of a uniform electron gas
with density )(rρ . The arrival of the overall expression for EXC involves the
multiplication of this term with the probability of finding an electron at that position
in space – the electron density )(rρ – as the weighting function.
))(( rXC ρε can be split into the exchange and correlation contributions, ))(( rX ρε and
))(( rC ρε respectively. The Slater exchange functional ))(( rX ρε is equal to:
31
)(343))(( ⎟
⎠⎞
⎜⎝⎛−=
πρρε rrX
(2-30)
Substituting equation (2-30) back to (2-29) yields the 34
ρ dependence of the exchange
energy.
Extension of the LDA to the unrestricted case leads to the LSDA. The exchange
energy for the LSDA is written as:
rdrrrrrE XCLSDAXC ))(),(()()](),([ βαβα ρρερρρ ∫= (2-31)
The LDA and LSDA are drastic approximations, since they assume that the density in
a molecular system is constant and can be approximated by a homogeneous uniform
electron gas.
2.11.2 Generalized Gradient Approximation (GGA)
The LDA and LSDA fail when the density undergoes rapid changes. An improvement
38
to the LDA and LSDA is the GGA, where the gradient of the electron density is taken
into account when computing the exchange energy, which can translate to an
improvement in accuracy of results. The exchange energy as a result of the GGA can
be written generically as:
∫ ∇∇= rdrrrrfrrE GGAXC ))(),(),(),(()](),([ βαβαβα ρρρρρρ (2-32)
Several parameterizations of the GGA exist today, such as the B88 [29] and the PW91
[30, 31] exchange functionals, some include the incorporation of empirical parameters,
and some are completely free of such. For instance, the B88 exchange functional
includes an empirical parameter – the exactly known exchange energies of the rare
gas atoms He through Rn, whereas the PBE functional [32, 33] is free of such
empirical parameters.
A similar approach has also been taken to obtain analytical forms of gradient
corrected correlation functionals, but their explicit forms are complicated and not
easily understood by simple physically motivated reasoning. A list of the explicit
forms of some exchange and correlation functionals is provided in Section A1.11 of
Appendix 1.
In principle, exchange functionals can be combined with correlation functionals to
calculate the exchange energy of a system. An example is the combination of
Becke’s functional with the correlation functional by Lee, Yang and Parr [34] – often
abbreviated as the BLYP functional.
39
2.11.3 Hybrid Functionals
Density based functionals often overestimate a quantity which HF theory
underestimates, for instance, bond lengths [35]. To address this, Becke suggested that
the exact exchange-correlation functional must include a fraction of HF exchange [36].
The exchange-correlation of the Kohn-Sham scheme is obtained through the
coupling-strength integrated exchange-correlation hole.
λλ dEE nclXC ∫=
1
0
(2-33)
The case when λ = 0 refers to an interaction free system, where the contribution to EXC
is the exchange energy only, and the correlation term is neglected. The case when λ =
1 refers to a fully interacting system, with exchange and correlation parts taken into
consideration in the computation of EXC.
The simplest case to consider is to assume that λnclE varies linearly with respect to λ.
This results in a half-and-half combination of exact exchange and density functional
exchange-correlation [37].
The most popular hybrid functional is Becke’s three parameter exchange functional
combined with the correlation functional of Lee, Yang and Parr (B3LYP). The
B3LYP exchange-correlation expression for the energy is as follows:
VWNC
LYPC
BXXC
DiracX
LYPBXC EccEbEaEEaE )1()1( 8803 −++++−= =λ (2-34)
40
The three parameters are optimized to be a = 0.20, b = 0.72 and c = 0.81.
Hybrid functionals such as B3LYP have achieved great success in the prediction of
molecular geometries and harmonic vibrational frequencies in recent years [38],
hence their widespread use in computational chemistry packages.
2.12 Composite Methods for Accurate Thermochemistry
From Figure 2-4, the exact solution of the Schrödinger equation is reached by using
full configuration interaction (FCI) and the complete basis set (CBS). However, this
task is not currently achievable, as the computing resources required (CPU, memory
and disk) is vast.
However, several composite methods have been proposed in recent years to calculate
accurate thermochemistry data. They usually incorporate a method for obtaining
accurate geometries compared to experiment, and an algorithm for extrapolating the
FCI/CBS energy. Two widely used composite methods are the Gaussian-n and CBS-n
methods developed by Pople et al and Petersson et al respectively.
2.12.1 Gaussian-n methods
The first Gaussian-n method, G1, was developed in 1989 by Pople et al [39]. The G1
method performed fairly well, yielding an accuracy of above 2 kcal mol–1 accuracy in
41
comparison with experimentally known atomization energies, ionization potentials,
and electron and proton affinities of a group of 31 molecules. In 1991, the second
generation of Gaussian-n methods was developed, namely the G2 method [40]. The
G2 method eliminated several deficiencies of G1 theory, and yielded an improvement
of 0.50 kcal mol–1 in mean absolute deviation from experimental heat of atomization
for 39 first row compounds. The use of reduced Møller-Plesset orders in calculation
of Gaussian-n energies was also introduced to reduce computation time and yet
minimize the compromise in accuracy [41].
The most current Gaussian-n method in use is the G3 family of methods, first
developed in 1998 [42]. Some key improvements of the G3 method over its
predecessors are the use of the G3Large basis set, addition of an empirical higher
level correction (HLC) term, and the consideration of core correlation effects. The
G3Large basis set is a modification of the 6-311+G(3df,2p) basis set used in G2, and
includes additional polarization functions for second row atoms (3d2f) and less for the
first row atoms (2df). Neutral basis sets were also used for the atoms P, S and Cl. The
HLC term is parameterized to obtain the least mean absolute deviation to that of
known experimental energies of the G2/97 test set [43, 44]. Core correlation effects
are included in a MP2(full)/G3Large energy calculation, which improves the accuracy
of calculated energies for systems with unsaturated ring systems.
The procedure for the G3 method is as follows: geometry optimization at the
MP2(full)/6-31G(d) level; zero point energies (ZPE) calculated at the HF/6-31G(d)
level, scaled by a factor of 0.8929; single point calculations carried out at the
42
QCISD(T)(fc)/6-31G(d), MP4(fc)/6-31+G(d), MP4(fc)/6-31G(2df,p), and
MP2(fu)/G3Large levels based on the MP2(fu)/6-31G(d) geometry.
The G3 energy for a molecule at 0 K is given as:
E0(G3) = E[MP4/6-31G(d)] + ΔE(+) + ΔE(2df) + ΔE(QCI)
+ ΔE(G3Large) + E(SO) + E(HLC) + E(ZPE) (2-35)
Where
ΔE(+) = E[MP4/6-31+G(d)] – E[MP4/6-31G(d)] (2-36)
ΔE(2df) = E[MP4/6-31G(2df,p)] – E[MP4/6-31G(d)] (2-37)
ΔE(QCI) = E[QCISD(T)(fc)/6-31G(d)] – E[MP4/6-
31G(d)] (2-38)
ΔE(G3Large)= E[MP2(fu)/G3Large]
– E[MP2(fc)/6-31G(2df,p)]
– E[MP2(fc)/6-31+G(d)]
+ E[MP2(fc)/6-31G(d)] (2-39)
E(HLC) = Anβ – B(nα – nβ) for molecules;
Cnβ – D(nα – nβ) for atoms (2-40)
E(SO) is the experimental or theoretically calculated spin-orbit correction. A, B, C
and D in E(HLC) are parameters having the values 6.386, 2.977, 6.219 and 1.185
millihartrees respectively.
Several variants of the G3 method are available as well. They include G3S, using
scaled energies for the HLC [45], G3X, using improved B3LYP/6-31G(2df,p)
43
geometry and ZPE [46], and G3-RAD, optimized for calculations of thermochemistry
of open-shell species [47]. G3 theory, together with its variants, can be carried out
with lower orders of Møller-Plesset perturbation [48], for example G3(MP2),
G3(MP3), G3X(MP2) and G3(MP2)-RAD. These are used for larger systems where
MP4 calculations are prohibitive, while minimizing loss of accuracy. Details of the
other Gaussian-n calculations used in this thesis are provided in section A2.12 of
Appendix 2.
2.12.2 CBS-n methods
The complete basis set (CBS) extrapolation methods were proposed by Petersson and
co-workers, who observed that basis set truncation was one of the main sources of
error in accurate calculation of molecular correlation energies. CBS extrapolations are
based on the asymptotic convergence of pair natural orbital (PNO) expansions of the
first order wave function [49].
∑=
=ΨN
CN1
)1( )(μ
μμϕ (2-41)
A graph of eij(N) versus N–1 for the helium atom (Figure 2.5) summarizes the general
idea of CBS extrapolation. The term eij(N) represents the correlation energy between
an electron pair, using N PNO expansions. The useful points for the extrapolation are
those corresponding to completely filled shells [50].
44
Figure 2.5. A graph of eij(N) versus (N + δ)–1 for the helium PNOs, adapted from reference 50.
The complete basis set–atomic pair natural orbital (CBS–APNO) method was
developed in 1993, and provided excellent accuracy (mean absolute error of 0.53 kcal
mol–1) for bond dissociation energies, ionization potentials and electron affinities for
the G2 test set [51]. The components of various CBS-n methods, with the components
of the G3 and G3(MP2) methods alongside are summarized in Table 2.1.
45
CBS-4 CBS-q CBS-Q CBS-QCI/APNO G3(MP2) G3 Geometry Method UHF/3-21G(*) UHF/3-21G(*) MP2(fc)/6-31G† QCISD/6-311G** MP2(fu)/6-31G* MP2(fu)/6-31G* ZPE Method UHF/3-21G(*) UHF/3-21G(*) UHF/6-31G† UHF/6-311G** UHF/6-31G* UHF/6-31G* Scale Factor 0.91671 0.91671 0.91844 0.9251 0.8929 0.8929 Core Correlation Method None None 3.92q(Na) +
2.83q2(Na) CBS2/KK,KL,LL,LL′ None None
SCF Method UHF UHF UHF UHF UHF UHF Basis sets H, He 31+G(p) 31+G(p) 311+G(2p) 4s2p1d 311+G(2p) 311+G(2p) Li – Ne 6-311+G(2df) 6-311+G(2df) 6-311+G(2df) 6s6p3d2f G3MP2Large G3Large Na – Ar 6-311+G(3d2f) 6-311+G(3d2f) 6-311+G(3d2f) N/A G3MP2Large G3Large MP2 Method CBS2 CBS2 CBS2 CBS2 MP2 MP2 Nmin 5 5 10 10 Npts 40 000 40 000 100 000 100 000 Basis sets H, He 31G† 31G† 311+G(2p) 4s2p1d 311+G(2p) 311+G(2p) Li – Ne 6-31+G† 6-31+G† 6-311+G(2df) 6s6p3d2f G3MP2Large G3Large Na – Ar 6-31+G† 6-31+G† 6-311+G(3d2f) N/A G3MP2Large G3Large MP4 Method MP4(SDQ) MP4(SDQ) MP4(SDQ) MP4(SDTQ) Basis sets 6-31G 6-31+G† 6-31+G(d(f),d,p) 6-31G* 6-31+G* 6-31G(2df,p) QCISD(T) Method QCISD(T) QCISD(T) QCISD(T) QCISD(T) QCISD(T) Basis sets 6-31G 6-31+G† 6-311G(d,p) 6-31G* 6-31G* Empirical corrections (mEh)
2 electron iiii
IS 252.5− iiii
IS 294.5− iiii
IS 233.5− iiii
IS 274.1− – 4.808npairs – 3.409npairs
1 electron – 4.55 (nα + nβ) – 4.03 (nα + nβ) 0 0 – 4.471 (nα + nβ) – 2.977 (nα + nβ) Spin – 38.43 Δ‹S2› 0 – 9.2 Δ‹S2› 0 0 0 Nmin = minimum number of natural orbital configurations used for CBS extrapolations Npts = number of mesh points per electron for the Monte Carlo numerical integration of |S|ij |S|ii = ∫|αφi
βφi|dτ , the absolute overlap integral Iii = (ΣCμ)2, the intraorbital interference factor
Table 2.1. Components of CBS-4, CBS-q, CBS-Q, CBS-QCI/APNO, G3(MP2) and G3 models. Adapted from Ref. 50.
2.13 Methods of Calculation of ESR G-shifts
G-tensors are an important part of any ESR spectrum. Calculations of EPR g-tensors
used in this thesis are based on DFT and the use of “gauge-including atomic orbitals”
(GIAO) schemes [52]. The g-tensor of EPR spectroscopy is considered as a second-
order property. The perturbation parameters are one Cartesian component of the
46
constant external magnetic field BS, and the net electronic spin component along a
given coordinate axis, St. The st tensor component of g is given by:
0
21
==∂∂
∂=
SBtSBst SB
Egμ
(2-42)
E is the total energy of the many-electron system. μB is the Bohr magneton.
The g-shift, Δg, is defined as the deviation of the molecular g-value from the free
electron value ge = 2.002 319 277 8.
ggg e Δ+= 1 (2-43)
1 is the unit tensor, g and gΔ are second-rank tensors. The isotropic g-shift, Δg is
the trace of gΔ . Contributions to Δg come from the operators hZ−KE, hSO−N, hSO−e,
hSOO, hSO−Ndia, hSO−e
dia, and hSOOdia, details of which are found in section A1.12 of
Appendix 1.
2.14 Solvation Models
Ab initio and DFT calculations are typically carried out for molecules in vacuo, to
determine their intrinsic properties. A great deal of experimental work is carried out
with reactants dissolved in a solvent, which typically changes the characteristics of
the molecules in question as compared to them in the gas phase. Solvent effects can
be modeled in the framework of self-consistent reaction field (SCRF) theory. A
47
macroscopic property of the solvent, such as the dielectric constant, is often used to
describe a continuous medium.
The solute is first placed in a cavity which is immersed in a continuous medium with
reaction field R . The solute is then treated quantum mechanically, and the
interactions with the solvent are taken into account by a perturbation on the solute
Hamiltonian.
Two continuum solvation models are used in this thesis. They are the Onsager
reaction field method [53, 54], and the integral equation formalism for the polarizable
continuum model (IEF-PCM) [55].
2.14.1 Onsager Reaction Field Theory
The solute is placed in a spherical cavity surrounded by a continuous medium of
dielectric constant, ε. A dipole in the solute induces a dipole in the solvent, and the
electric field experienced by the solute interacts with the molecular dipole.
The electrostatic solvent effect may be taken as a perturbation in the Hamiltonian of
the isolated molecule. This perturbation term describes the coupling between the
molecular dipole operator and the reaction field, R .
RHH fieldrxn ⋅−=− μ0 (2-44)
48
The reaction field R is proportional to the molecular dipole moment, μ , hence
μgR = , where the proportionality constant, g is related to ε and the radius of the
spherical cavity, a0.
3
0)12()1(2
ag
+−
=εε
(2-45)
For an SCF wave function, the effects of R are incorporated into the Fock matrix as
an additional term.
⟩⟨−= σλλσλσ ϕμϕμ |ˆ|0 gFF (2-46)
The energy of the system is hence given by
RHE ⋅−⟩ΨΨ⟨= μ21
0 || (2-47)
In the case of a molecular ion with a total charge Q, a Born charge term is added to
the energy expression, to account for the ion-dipole interaction. The quantities E, Q
and a0 are in atomic units, namely Hartree, units of electron charge, and Bohr radii,
respectively.
0
2
)(11
21
aQEE ion ⎟
⎠⎞
⎜⎝⎛ −−=
ε
(2-48)
2.14.2 Integral equation formalism for polarizable continuum solvent model
(IEFPCM)
The IEFPCM procedure considers an infinitely dilute solution at equilibrium [56].
The reference state of a solvent at a given temperature T, supplemented by a number
49
of non interacting electrons and nuclei sufficient to build the solute M, is used to
consider the Gibbs free energy of solvation. A cavity is constructed around M, formed
by interlocking spheres, centred on solute atoms or atomic groups. Several sets of
atomic radii, such as the united atom topological model (UATM) [57] can be used to
create such a cavity. The work spent in constructing the cavity in the solvent
corresponds to Gcav, the cavitation free energy. Free electrons and nuclei are then
collected in the cavity to form M. The work spent in this second process gives a
second contribution to the free energy of the system. Phenomenological partitioning
of these free energies gives:
cavMmdisrepel
S GGGGGG ++++= (2-49)
Where GS is the difference of free energy with respect to the reference state, Gel, Grep,
Gdis and GMm are the free energy contributions in solution due to electrostatic
interactions, repulsion, dispersion and molecular motion respectively. The Gibbs free
energy change of solvation would then correspond to the difference between GS and
the Gibbs free energy of M in the gaseous state, i.e.
0GGG Ssol −=Δ (2-50)
And hence:
Mmcavdisrepelsol GGGGGG Δ++++Δ=Δ (2-51)
The interaction potential can also be decomposed as:
repdisel VVVV ++=int (2-52)
50
The quantum mechanical problem can be expressed in terms of the following
Schrödinger equation:
)()())(( int0 xExVH M Ψ=Ψ+ ρ (2-53)
Vint depends on the charge distribution of M. In standard PCM, the Vdis and Vrep terms
are eliminated, and corresponding GS is obtained with the aid of empirical expressions
which do not exhibit explicit dependence on ρM. The framework of the Born-
Oppenheimer approximation is still used, and thus only the electronic component of Ψ
in equation (2-60) is considered. The linear dependence of Vel is reduced to the ρeM
term, and Vint is partitioned into two components:
intintint )( VVV e
M ′′+′= ρ (2-54)
where *)(int ΨΨ=′′ AV (2-55)
A is an appropriate integral operator. The free energy functional JS subject to
minimization is written as:
⟩Ψ′+′′+Ψ⟨= |)(ˆ
21ˆˆ| intint
0 eM
S VVHJ ρ(2-56)
In the Hartree-Fock case, JS is reduced to the matrix equation
F′C=SCε (2-57)
The Fock matrix here includes terms accounting for the presence of the solvent field,
i.e.
F′= h + ½(j+y) + G(P) + X(P) (2-58)
51
The matrices h and G(P) contain the one and two-electron integrals collected in the
SCF calculations in vacuo, P is the one-electron density matrix, j, y, and X(P) collect
the one- and two-electron integrals added when the interactions with the polarized
dielectric medium are explicitly taken into account within the PCM framework [55].
The final output is an electronic wave function from which properties can be derived,
together with a free energy of solvation.
References
1. Hehre, W. J.; Radom, L.; Schleyer, P. v.R.; Pople, J. A. Ab Initio Molecular
Orbital Theory; Wiley: New York, NY, 1986.
2. Foresman, J. B.; Frisch, Æ. Exploring Chemistry with Electronic Structure
Methods; 2nd Ed.; Gaussian Inc.: Pittsburgh, PA, 1996.
3. Lowe, J. P. Quantum Chemistry; 2nd Ed.; Academic Press: San Diego, CA,
1993.
4. Schrödinger, E. Ann. Phys. 1926, 79, 361.
5. Schrödinger, E. Phys. Rev. 1926, 28, 1049.
6. Hartree, D. R. Proc. Cambridge Phil. Soc. 1927, 24, 111.
7. McWeeny, R. Nature 1973, 243, 196.
8. Born, M.; Oppenheimer, R. Ann. Phys. 1927, 84, 457.
9. Hartree, D. R. Proc. Cambridge Phil. Soc. 1928, 24, 426.
10. Fock, V. Z. Phys. 1930, 61, 126.
11. Roothaan, C. C. J. Rev. Mod. Phys. 1951, 23, 69.
52
12. Hall, G. G. Proc. Roy. Soc. 1951, A205, 541.
13. Pople, J. A.; Nesbet, R. K. J. Chem. Phys. 1954, 22, 571.
14. Pople, J. A.; Head-Gordon, M.; Raghavachari, K. J. Chem. Phys. 1987, 87,
5968.
15. Purvis III, G. D.; Bartlett, R. J. J. Chem. Phys. 1982, 76, 1910.
16. Urban, M.; Noga, J.; Cole, S. J.; Bartlett, R. J. J. Chem. Phys. 1985, 83, 4041.
17. Noga, J.; Bartlett, R. J. J. Chem. Phys. 1987, 86, 7041.
18. Raghavachari, K.; Trucks, G. W.; Pople, J. A.; Head-Gordon, M. Chem. Phys.
Lett. 1989, 157, 479.
19. Møller, C.; Plesset, M. S. Phys. Rev. 1934, 46, 618.
20. Hinze, J. J. Chem. Phys. 1973, 59, 6424.
21. Roos, B. O.; Taylor, P. R. Chem. Phys. 1980, 48, 157.
22. Siegbahn, P. E. M.; Almlöf, J.; Heiberg, A.; Roos, B. O. J. Chem. Phys. 1981,
74, 2384.
23. Siegbahn, P. E. M.; Heiberg, A.; Roos, B. O., Levy, B. Phys. Scripta 1980, 21,
323.
24. Kohn, W.; Becke, A. D.; Parr, R. G. J. Phys. Chem. 1996, 100, 12974.
25. Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and Molecules;
Oxford University Press: Oxford, 1989.
26. Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density Functional Theory;
Wiley: Weinheim, 2000.
27. Hohenberg, P.; Kohn, W. Phys. Rev. 1964, 136, B864.
28. Kohn, W.; Sham, L. J. Phys. Rev. 1965, 140, A1133.
29. Becke, A. D.; Phys. Rev. A, 1988, 38, 3098.
30. Perdew, J. P., Wang, Y.; Phys. Rev. B 1992, 45, 13244.
53
31. Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.;
Singh, D. J.; Fiolhais, C.; Phys. Rev. B 1992, 46, 6671.
32. Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865.
33. Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1997, 78, 1396.
34. Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785.
35. Johnson, B. G.; Gill, P. M. W.; Pople, J. A. J. Chem. Phys. 1993, 98, 5612.
36. Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
37. Becke, A. D. J. Chem. Phys. 1993, 98, 1372.
38. Raghavachari, K. Theor. Chem. Acc. 2000, 103, 361.
39. Pople, J. A.; Head-Gordon, M.; Fox, D. J. J. Chem. Phys. 1989, 90, 5622.
40. Curtiss, L. A.; Raghavachari, K.; Trucks, G. W.; Pople, J. A. J. Chem. Phys.
1991, 94, 7221.
41. Curtiss, L. A.; Raghavachari, K.; Pople, J. A. J. Chem. Phys. 1993, 98, 1293.
42. Curtiss, L. A.; Raghavachari, K.; Redfern, P. C.; Rassolov, V.; Pople, J. A. J.
Chem. Phys. 1998, 109, 7764.
43. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Pople, J. A. J. Chem. Phys.
1997, 106, 1063.
44. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Pople, J. A. J. Chem. Phys.
1998, 109, 42.
45. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Pople, J. A. J. Chem. Phys.
2000, 112, 1125.
46. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Pople, J. A. J. Chem. Phys.
2001, 114, 108.
47. Henry, D. J.; Sullivan, M. B.; Radom, L. J. Chem. Phys. 2003, 118, 4849.
54
48. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Rassolov, V.; Pople, J. A. J.
Chem. Phys. 1999, 110, 4703.
49. Nyden, M. R.; Petersson G. A. J. Chem. Phys. 1981, 75, 1843.
50. Ochterski, J. W.; Petersson, G. A.; Montgomery, Jr. J. A. J. Chem. Phys. 1996,
104, 2598.
51. Ochterski, J. W.; Petersson, G. A.; Montgomery, Jr. J. A. J. Chem. Phys. 1994,
101, 5900.
52. Schreckenbach, G.; Ziegler, T. J. Phys. Chem. A 1997, 101, 3388.
53. Wong, M. W.; Frisch, M. J.; Wiberg, K. B. J. Am. Chem. Soc. 1991, 113, 4776.
54. Wong, M. W.; Wiberg, K. B.; Frisch, M. J. J. Chem. Phys. 1991, 95, 8991.
55. Cancès, E.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107, 3032.
56. Tomasi, J.; Mennucci, B.; Cancès, E. J. Mol. Struct. (THEOCHEM) 1999, 464,
211.
57. Barone, V.; Cossi, M.; Tomasi, J. J. Chem. Phys. 1997, 107, 3210.
55
Chapter Three
Thermochemistry of Reactive Nitrogen Oxide Species
3.1 Introduction
Reactive nitrogen oxide species (RNOS) are of great importance in many areas of
chemistry and biology. For instance, nitric oxide (NO•) is known to have contrastive
roles. Its important effects range from cytoprotective, to regulatory, and even to
cytotoxic [1– 4]. There is also strong interest in the spectrum of reactions of RNOS.
For instance, the RNOS peroxynitrite (ONOO–) and peroxynitrous acid (ONOOH) are
involved in atmospheric chemistry [5] and biochemical processes [6, 7]. They are
capable of undergoing a vast number of biological reactions, resulting in the
formation of other RNOS, which in turn have direct or indirect effects in many
important biochemical processes.
RNOS, being highly reactive, are transient in nature. Hence it is not easy to ascertain
their chemical properties, such as their standard enthalpies of formation, and
molecular structures, experimentally. To better understand the thermochemistry of
these RNOS, we have performed high-level ab initio molecular orbital calculations to
benchmark the suitability of various theoretical methods, and also to provide more
definitive estimates of the heats of formation of ONOO– and ONOOH, for which
experimental thermochemical data are lacking.
56
To this end, a series of sixteen biologically relevant RNOS, namely NO•, NO+, NO–,
HNO, NO2•, NO2
+, NO2–, HNO2, N2O, NO3
•, ONOO–, ONOOH, NOCl, NO2Cl, N2O3
and N2O4 (see Figure 3-1) were studied using various popular composite methods.
These are the Gaussian-3 methods and their variants (G3(MP2), G3X(MP2), G3 and
G3X) and the complete basis set (CBS) extrapolation methods (CBS-4, CBS-q, CBS-
Q, CBS-QB3 and CBS-APNO), which are well established for giving reliable
thermochemistry. The standard enthalpies of formation (ΔH°f) at 298 K were
calculated. In addition, the 298 K reaction enthalpies of three reactions related to the
decomposition of ONOO– and ONOOH (eqs. (1)–(3)) were also examined using the
CBS-RAD and G3-RAD methods of Radom and co-workers. These RAD methods
were designed to provide better theoretical treatment for free radicals by using
restricted open-shell (ROHF) wave functions in place of unrestricted (UHF) wave
functions, thus minimizing the effects of unwanted spin contamination from higher
electronic states in UHF formalisms.
ONOO– → NO• + O2•– (1)
ONOO– → NO2• + O2
•– (2)
ONOOH → NO2• + OH• (3)
57
N1 O 1
NO•2ΠC∞v
N1 O 1
NO+1Σ+
C∞v
N1 O1
NO–3Σ–
C∞v
N1 O1
H
HNO1A′Cs
N1 O1
O2
NO2•
2A1C2v
N1 O1O2
NO2+
1Σg+
D∞h
N1 O1
O2
NO2–
1A1C2v
HO1
O2 N1
HNO21A′Cs
N1 N2O1
N2O1Σg
+
C∞v
N1 O1
O2
O3
NO3•
2A2′D3h
N1 O1
O2
O3
N1 O1
O2
O3
H
N1 O1
Cl 1
N1 O1
O2
Cl 1
N1 O1
O2
N2
O3
O4
ONOO–1A′Cs
ONOOH1A′Cs
N1 O1
N2O2
O3
NOCl1A′Cs
NO2Cl1A1C2v
N2O31A′Cs
N2O41AgD2h
N1 O 1
NO•2ΠC∞v
N1 O 1
NO+1Σ+
C∞v
N1 O1
NO–3Σ–
C∞v
N1 O1
H
HNO1A′Cs
N1 O1
O2
NO2•
2A1C2v
N1 O1O2
NO2+
1Σg+
D∞h
N1 O1
O2
NO2–
1A1C2v
HO1
O2 N1
HNO21A′Cs
N1 N2O1
N2O1Σg
+
C∞v
N1 O1
O2
O3
NO3•
2A2′D3h
N1 O1
O2
O3
N1 O1
O2
O3
H
N1 O1
Cl 1
N1 O1
O2
Cl 1
N1 O1
O2
N2
O3
O4
ONOO–1A′Cs
ONOOH1A′Cs
N1 O1
N2O2
O3
NOCl1A′Cs
NO2Cl1A1C2v
N2O31A′Cs
N2O41AgD2h
Figure 3.1. Structures, element numbering, state and symmetry symbols of the sixteen RNOS in this study.
3.2 Computational Methods
The standard enthalpies of formation at 298 K (ΔH°f,298) of the sixteen RNOS
mentioned earlier were studied using the following Gaussian-3 and CBS methods:
58
G3(MP2) [8], G3X(MP2) [9], G3 [10], G3X [9], CBS-4 [11, 12], CBS-q [13], CBS-Q
[11], CBS-QB3 [14] and CBS-APNO [15]. These composite methods are designed to
predict reliable thermochemistry, and are benchmarked with known experimental
energies in the G2/97 test set. The overall mean absolute deviations (MAD) of
G3(MP2), G3X(MP2), G3 and G3X theories from experiment for 299 energies in the
G2/97 test set are 1.30 [8], 1.25 [9], 1.02 [9] and 0.96 kcal mol–1 [10] respectively.
For the CBS methods, MAD for the CBS-4 method is 3.26 kcal mol–1 on the G2/97
test set [12]. The benchmarks for the CBS-q, CBS-Q, CBS-QB3 and CBS-APNO
methods were 125 energies of the G2 test set for CBS-q and CBS-Q, the entire G2 test
set for CBS-QB3, and 64 energies involving only first row atoms from the G2 test set
for CBS-APNO. The reported MAD for these methods are 1.71 [11], 1.01 [11], 0.87
[14] and 0.53 kcal mol–1 [15] for the CBS-q, CBS-Q, CBS-QB3 and CBS-APNO
methods respectively. The low values of MAD with experimental values for these
composite methods gave confidence that this was the method of choice to predict
ΔH°f,298 for those species without known experimental values, namely NO2
+, ONOO–,
and ONOOH.
Reaction enthalpies at 298 K of reactions (1)–(3) were also examined by CBS-RAD
[16], G3(MP2)-RAD [17], G3X(MP2)-RAD [17], G3(MP2), G3X(MP2) and
CCSD(T) theories. The standard enthalpies of formation were computed using the
atomization approach of Radom and co-workers [18]. This method involves the use of
experimental heats of formation of gaseous atoms and ions at 0 K, obtained from
NIST-JANAF thermochemical tables [19]. For instance, the ΔH°f,298 of nitric oxide
(NO•) is computed as follows:
59
ΔH°f,0 (NO•) = H°0 (NO•) – H°0 (N(g)) – H°0 (O(g))
+ ΔH°f,0 (N(g)) + ΔH°
f,0 (O(g))
ΔH°f,298 (NO•) = ΔH°
f,0 (NO•) + [H°f,298 (NO•) – H°0 (NO•)]
– [H°f,298 (N(g)) – H°0 (N(g))]
– [H°f,298 (O(g)) – H°0 (O(g))]
The cis conformation of ONOO–, and the (cis,cis)-planar form of ONOOH were
employed in this study, as they were shown to be the preferred conformations based
on theoretical calculations and NMR studies [20]. For ONOO–, the cis conformation
is in accord with the observed geometry from the crystal structure of tetraammonium
peroxynitrite [21].
Full valence complete-active-space SCF (CASSCF) geometry optimizations were also
carried out for ONOO–. URCCSD(T) calculations employed in the G3(MP2)-RAD
and G3X(MP2)-RAD methods, as well as the CASSCF calculations were performed
with the MOLPRO 2002 [22] program. All other calculations were carried out using
the GAUSSIAN 98 [23] suite of programs.
3.3 Results and Discussion
3.3.1 Enthalpies of Formation of RNOS
The calculated enthalpies of formation (ΔH°f,298) of the sixteen RNOS are presented in
Tables 3.1 and 3.2. There is fair to excellent agreement of the calculated values
60
compared with the known experimental values [19, 24], with mean average deviations
(MAD) ranging from 5.1 to 12.9 kJ mol–1. However, we note that the MAD values are
somewhat larger than the corresponding values reported originally for a larger set of
molecules, the G2/97 test set. Among the various composite methods considered here,
the G3X(MP2) method has the smallest MAD of 4.9 kJ mol–1, and smallest maximum
deviation (MD) of 12.0 kJ mol–1. The G3 method performs marginally better, with a
MAD and MD of 5.1 and 12.9 kJ mol–1 respectively. The Gaussian-n methods
generally perform somewhat better than the CBS methods.
Species G3(MP2) G3 G3X(MP2) G3X Expt. NO• 88.0 87.5 85.9 86.8 90.29 ± 0.17 NO+ 999.1 997.6 987.9 985.8 990.19 ± 0.8 NO– 76.7 75.9 76.7 76.4 88.70 ± 1.3 HNO 105.0 104.8 100.4 100.2 99.58 NO2
• 32.6 28.5 29.3 25.3 33.10 ± 0.8 NO2
+ 951.9 958.6 956.9 946.9 NO2
– –200.8 –203.4 –203.1 –205.9 –202.72 ± 5.8 HNO2 –77.7 –80.1 –82.6 –83.9 –76.73 ± 1.34 N2O 86.8 83.8 77.2 73.6 82.05 ± 0.4 NO3
• 94.9 84.0 76.2 65.6 71.13 ± 20.9 ONOO– –103.8 –108.2 –108.0 –112.3 ONOOH –12.6 –17.6 –17.7 –22.7 NOCl 48.5 56.9 47.5 49.0 51.71 ± 0.42 NO2Cl 8.4 9.4 6.6 1.3 12.13 ± 1.7 N2O3 88.7 84.8 75.5 70.8 82.84 ± 0.8 N2O4 14.0 3.6 2.4 –8.9 9.08 ± 1.7 Mean abs. dev.
6.0 5.1 4.9 7.4
Max. dev. 23.8 12.9 12.0 18.0 Table 3.1. Calculated standard enthalpies of formation (ΔH°
f,298, kJ mol–1) of sixteen RNOS using various Gaussian-n methods.
61
Species CBS-4 CBS-q CBS-Q CBS-QB3 CBS-APNO
Expt.
NO• 108.3 79.9 84.6 83.1 88.5 90.29 ± 0.17 NO+ 998.9 987.7 988.4 985.3 984.2 990.19 ± 0.8 NO– 83.7 84.3 73.5 73.7 80.0 88.70 ± 1.3 HNO 107.4 103.0 100.5 97.2 103.6 99.58 NO2
• 24.1 6.6 18.0 19.6 21.2 33.10 ± 0.8 NO2
+ 968.7 939.9 942.5 932.3 938.8 NO2
– –191.5 –205.1 –205.6 –208.4 –196.0 –202.72 ± 5.8 HNO2 –73.3 –87.8 –89.0 –90.3 –83.9 –76.73 ± 1.34 N2O 99.2 82.9 74.3 68.7 71.6 82.05 ± 0.4 NO3
• 78.4 35.5 63.7 51.1 69.6 71.13 ± 20.9 ONOO– –50.4 –87.3 –114.9 –115.0 –37.0 ONOOH 9.1 –17.5 –31.1 –31.8 –25.1 NOCl 52.0 46.6 40.1 43.6 – 51.71 ± 0.42 NO2Cl 3.0 –12.3 –10.3 –8.5 – 12.13 ± 1.7 N2O3 83.9 58.7 68.2 66.0 75.1 82.84 ± 0.8 N2O4 –0.8 –32.0 –13.9 –17.2 –7.3 9.08 ± 1.7 Mean abs. dev.
8.3 14.8 10.8 12.9 7.5
Max. dev.
18.0 41.1 23.0 26.3 16.4
Table 3.2. Calculated standard enthalpies of formation (ΔH°
f,298, kJ mol–1) of sixteen RNOS using various CBS methods. For the three RNOS systems without known experimental heats of formation, namely
NO2+, ONOO–, and ONOOH, the agreement between G3 and CBS-4, CBS-q and
CBS-APNO values is rather poor. For instance, the predicted heats of formation of
ONOO– by the G3 and CBS-APNO methods differ greatly, by 71 kJ mol–1! Other
methods predict values (–104 to –115 kJ mol–1) close to that of the G3 result (–108 kJ
mol–1). A careful inspection of the geometries employed by the G3 and CBS-APNO
methods reveals significant differences in the central N—O bond length in ONOO–,
which may be a possible cause in the error in the calculated heat of formation of
ONOO–.
Selected structural parameters of the sixteen RNOS obtained by the various composite
methods are tabulated in Table 3.3. Included are CCSD(T)/cc-pVTZ and experimental
geometrical parameters [19] for comparison. It is observed that the experimental N—
62
O and N—N bond lengths of RNOS are readily produced by B3LYP and MP2
methods, with B3LYP giving better agreement than MP2. This is in accord with
previous findings based on a comparative study of the geometries of nitrogen oxides
by Bohr and co-workers [5].
Species Parameter HF/ a CBS-4 CBS-q
B3LYP/ b CBS-QB3
B3LYP/ c G3X
MP2(fu)/ d G3
MP2(fc)/ e CBS-Q
QCISD/ f CBS-APNO
CCSD(T)/ g
Expt.
NO• r(N1—O1) 1.201 1.148 1.151 1.143 1.137 1.159 1.153 1.151 NO+ r(N1—O1) 1.047 1.060 1.064 1.103 1.090 1.067 1.069 NO– r(N1—O1) 1.312 1.273 1.272 1.280 1.270 1.275 1.275 HNO r(N1—O1) 1.216 1.200 1.201 1.236 1.223 1.207 1.214 1.211 NO2
• r(N1—O1) = r(N1—O2)
1.204 1.195 1.196 1.165 1.205 1.194 1.199 1.193
NO2+ r(N1—O1)
= r(N1—O2) 1.137 1.137 1.121 1.091 1.137 1.137 1.126
NO2– r(N1—O1)
= r(N1—O2) 1.287 1.265 1.263 1.281 1.271 1.261 1.266
HNO2 r(N1—O1) r(N1—O2)
1.190 1.420
1.181 1.388
1.184 1.382
1.209 1.385
1.198 1.376
1.184 1.379
1.187 1.392
1.169 1.442
N2O r(N1—N2) r(N1—O1)
1.095 1.260
1.126 1.184
1.128 1.186
1.092 1.179
1.169 1.182
1.129 1.189
1.133 1.190
1.128 1.184
NO3• r(N1—O1)
=r(N1—O2) =r(N1—O3)
1.245 1.233 1.233 1.255 1.246 1.230 1.237
ONOO– r(N1—O1) r(N1—O2) r(O2—O3)
1.227 1.343 1.533
1.218 1.370 1.382
1.221 1.365 1.370
1.188 1.270 1.428
1.239 1.340 1.358
1.214 1.721 1.292
1.222 1.361 1.389
1.16 1.35 1.41
ONOOH r(N1—O1) r(N1—O2) r(O2—O3)
1.184 1.447 1.463
1.184 1.390 1.429
1.189 1.379 1.420
1.211 1.385 1.446
1.197 1.384 1.433
1.184 1.390 1.419
1.199 1.402 1.438
NOCl r(N1—O1) r(N1—Cl1)
1.148 1.907
1.126 2.044
1.134 1.990
1.156 2.031
1.141 2.057
– –
1.141 1.982
1.139 1.975
NO2Cl r(N1—O1) =r(N1—O2) r(N1—Cl1)
1.221
1.762
1.184
1.939
1.188
1.903
1.206
1.906
1.192
1.943
–
–
1.205
1.825
1.202
1.840 N2O3 r(N1—N2)
r(N1—O1) r(N2—O2) r(N2—O3)
1.671 1.153 1.217 1.220
1.879 1.135 1.200 1.201
1.856 1.140 1.201 1.202
1.935 1.173 1.220 1.222
1.929 1.161 1.209 1.211
1.751 1.148 1.200 1.201
1.836 1.147 1.205 1.205
1.864 1.142 1.217 1.202
N2O4 r(N1—N2) r(N1—O1) =r(N1—O2) =r(N2—O3) =r(N2—O4)
1.637 1.204
1.798 1.188
1.780 1.190
1.820 1.212
1.828 1.202
1.688 1.190
1.744 1.194
1.782 1.190
Table 3.3. Selected calculated bond lengths of RNOS in Ångstroms. Basis sets: (a) 3-21G(d), (b) 6-311G(2d,d,p), (c) 6-31G(2df,p), (d) 6-31G(d), (e) 6-31G†, (f) 6-311G(d,p), (g) cc-pVTZ.
The QCISD bond lengths are in good agreement with the CCSD(T)/cc-pVTZ results
and experimental values, with a few notable exceptions. The N—O bond length in
ONOO– and the N—N bond length in N2O3 and N2O4. For ONOO–, QCISD theory
63
predicts a rather unexpectedly long bond length of 1.72–1.73 Å. MP2, B3LYP and
CCSD(T) methods yield a typical N—O bond length of 1.340–1.375 Å. For the N—N
bonds of N2O3 and N2O4, QCISD bond distances are too short by 0.1 Å compared to
the experimental values.
To further shed light on this intriguing discrepancy in molecular geometries, the
structure of ONOO– was optimized using various single-reference and multi-reference
based methods with the 6-31G(d) basis set. The default convergence criteria for
geometry optimization in the GAUSSIAN98 and MOLPRO2002 software packages
were employed. For the GAUSSIAN98 program, the values of maximum force, root-
mean square (RMS) force, maximum step, and RMS step have to fall within threshold
values of 0.000450, 0.000300, 0.001800 and 0.001200 a.u., respectively, to achieve
convergence. For the MOLPRO2002 program, the values for maximum force and
maximum step have to be less than 0.000300 a.u. to achieve convergence. The results
are summarized in Table 3.4. Most methods predict a ‘normal’ N—O bond distance
of 1.30–1.40 Å except for QCISD. The closely related CCSD method yields an
unexpectedly different geometry when compared to QCISD. Somewhat longer N—O
bond lengths are also obtained by pure functionals SVWN and BLYP in density
functional theory. It is interesting to note that inclusion of triple excitations to QCISD
(i.e. QCISD(T)) gives good geometry, compared to the higher-level results. A similar
trend is obtained for the larger cc-pVTZ basis set (see Table 3.4), indicating that the
QCISD problem for ONOO– is independent of the basis set used. The best calculated
N—O bond lengths at the CCSD(T)/cc-pVTZ and full-valence-CASSCF/cc-pVTZ
levels are 1.361 and 1.358 Å respectively, in excellent accord with the experimental
value of 1.35 Å, derived from the X-ray structure of tetraammonium peroxynitrite
64
[21]. To the best of our knowledge, we are unaware of any example where QCISD
and CCSD theories yield significantly different molecular geometries. It is worth
noting that there are problematic closed-shell systems such as O22+ [25] which exhibit
slow convergence of the Møller-Plesset perturbation theory. However, the MPn series
appears to converge normally for ONOO–.
Level r(N1—O1)
r(N1—O2) r(O2—O3) α(O1N1O2) α(N1O2O3)
6-31G(d) basis set
HF 1.188 1.269 1.428 118.5 118.3 SVWN 1.222 1.428 1.330 112.6 115.9 BLYP 1.234 1.541 1.367 112.3 115.6 B3LYP 1.227 1.374 1.377 115.3 117.9 MP2 1.251 1.348 1.370 114.9 117.6 MP3 1.220 1.314 1.404 116.5 117.3 MP4 1.261 1.391 1.361 113.9 117.5 BD 1.226 1.330 1.414 116.5 117.2 CISD 1.211 1.301 1.407 117.3 117.6 QCISD 1.232 1.729 1.306 104.1 108.4 QCISD(T) 1.236 1.397 1.402 114.8 116.8 CCSD 1.227 1.340 1.411 116.2 117.1 CCSD(T) 1.238 1.383 1.397 114.9 116.9 CAS(24,16) 1.229 1.367 1.437 116.3 115.6 cc-pVTZ basis set HF 1.178 1.264 1.415 118.8 118.5 B3LYP 1.215 1.361 1.385 116.2 118.3 MP2 1.234 1.336 1.361 115.4 117.9 QCISD 1.210 1.709 1.296 105.2 109.5 QCISD(T) 1.209 1.383 1.371 115.2 116.5 CCSD 1.210 1.322 1.400 116.9 117.5 CCSD(T) 1.222 1.361 1.389 115.8 117.4 CAS(24,16) 1.219 1.358 1.432 116.8 117.2 Table 3.4. Calculated structural parameters of planar cis-peroxynitrite, ONOO–, at various levels of theory.
From the above discussion, it is clear that the significant difference in CBS-APNO
heat of formation of ONOO– compared with experiment and with other composite
methods, is due to inadequate QCISD geometry. Hence, the best predicted heats of
formation of ONOO–, NO2+, and ONOOH are the G3 values (Table 3.2).
65
The exothermicity of formation for ONOO– (–108 kJ mol–1) and ONOOH (–18 kJ
mol–1) at 298 K, indicates the stabilities of these compounds with respect to their
constituent elements in their standard states. It must be noted, however, that these
species are considerably less stable than the commonly found nitrate ion (NO3–) and
nitric acid (HNO3). The G3 relative energies of NO3– and HNO3 are –219.3 and –
122.3 kJ mol–1 respectively. Both species are predicted to have a sizable activation
barrier at the G3 level for the exothermic unimolecular rearrangement, 242.6 and
208.9 kJ mol–1 for ONOO– and ONOOH respectively. Although these unimolecular
rearrangements are unlikely due to their prohibitively high barriers, these processes
are catalyzed by proteins such as human hemoglobin, within a biological system [26].
The predicted proton affinity for ONOO– at 298 K is 1446.5 kJ mol–1 at the G3 level,
which is consistent with the value reported by Klasinc and co-workers [27].
Although the G3 method shows promising results in predicting heats of formation, it
should be noted that significant deviations occur for open shell species NO– (3Σ–) and
NO3• (2A2′) (–12.8 and +12.9 kJ mol–1 respectively). The use of UHF wave functions
in G3 theory on these species may be the possible source of error in the calculated
enthalpies. For NO3•, the discrepancy between calculated and experimental energies
can also be explained in terms of the pseudo Jahn-Teller effect. There are strong
interactions between its ground state (2A2′) and the second excited state (2E′), as
explored by Janoschek and co-workers [28].
Interestingly, the G3X heats of formation are found to be less accurate than the
corresponding G3 values, with G3X having a MAD of 7.4 kJ mol–1 compared to 5.1
kJ mol–1 for G3. This finding is somewhat surprising, given the fact that B3LYP
66
geometries are more reliable than MP2 geometries for this set of RNOS. Perhaps the
high level correction (HLC) term for G3X theory may not be optimum for the RNOS
studied here.
It is important to note that there are two ways of computing atomization energy for
the ionic species. For instance, the two possible reaction channels of NO – correspond
to N + O– and N– + O. For all the anions, the two approaches give rise to two sets of
significantly different heats of formation. By comparing with well established heats of
formation of nitrogen-containing compounds, it is found that the heat of formation of
the nitrogen anion (N–) cited in the JANAF tables is unreliable. The reported value of
473.5 kJ mol–1 is likely to be underestimated by about 15 kJ mol–1.
3.3.2 Enthalpies of reactions involving ONOO– and ONOOH decomposition
The calculated reaction enthalpies for the decomposition reactions of ONOO– and
ONOOH (reactions (1) – (3)) are presented in Table 3.5. The results, using various
levels of theory, are generally in close agreement with each other, typically within
~10 kJ mol–1.
Notable exceptions are the enthalpies of reactions (1) and (2) calculated using the
CBS-RAD and CBS-APNO methods. The computed enthalpies using these two
methods are considerably different from other calculated values, differing by up to 75
kJ mol–1. Both CBS-RAD and CBS-APNO methods are based on QCISD geometry,
in conjunction with the 6-31G(d) and 6-311G(d,p) basis sets respectively. As shown
in the previous section, the geometry of ONOO– is poorly described by QCISD theory.
67
Hence, the large enthalpy deviation using these methods can be attributed to the use
of inadequate QCISD geometries. This is readily confirmed by performing modified
CBS-RAD and CBS-APNO calculations, where CCSD geometries are employed
instead of QCISD geometries. The revised reaction enthalpies (Table 3.5) derived
from the CCSD geometry are now in good agreement with other computed values. A
comparison of these results for reaction enthalpies with those reported by Olson et al
[29] and Pfeiffer et al [30] shows consistency with calculated values for CBS-QB3
and G3(MP2)//B3 methods. Interestingly, the geometry of the protonated form of
ONOO− (i.e. ONOOH) is well described by QCISD theory (Table 3.3). As a
consequence, the computed reaction enthalpies by various methods for reaction (3)
agree well with each other. The DFT method B3LYP/aug-cc-pVTZ//B3LYP/cc-
pVDZ predicts a fairly large deviation of 20% from the G3 enthalpy in all cases
(Table 3.5).
Method Reaction (1) enthalpy Reaction (2) enthalpy Reaction (3) enthalpy G3(MP2) 147.0 242.2 76.2 G3X(MP2) 148.7 243.2 76.8 CBS-Q 141.9 238.8 82.8 G3(MP2)-RAD 135.6 244.4 79.1 G3X(MP2)-RAD 136.1 243.4 77.9 CBS-RAD 96.8 (136.4) 199.1 (239.8) 84.7 CBS-APNO 75.9 (142.7) 164.0 (231.7) 80.5 G3 146.2 236.4 77.5 G3X 149.1 243.4 78.3 B3LYP/aug-cc-pVTZ/ /B3LYP/cc-pVDZ
124.1 209.1 62.6
CCSD(T)/aug-cc-pVTZ// CCSD/cc-pVTZ
129.8 225.9 82.7
CCSD(T)/cc-pVQZ// CCSD/cc-pVTZ
138.3 258.8 83.9
Table 3.5. Calculated reaction enthalpies, in kJ mol–1, for reactions (1) to (3) at 298 K. Values in parentheses indicate reaction enthalpies calculated based on CCSD instead of QCISD geometries for CBS-RAD and CBS-APNO methods.
Reactions involving the decomposition of ONOO− and ONOOH via various pathways
have attracted strong interest due to their biological importance, and their
68
thermochemistry is a subject of intense debate among various research groups. For
instance, the harmful effects of ONOOH has been attributed to the apparent release of
NO2• and OH• radicals upon unimolecular decomposition and the direct reaction of
these radicals on other substrates [31]. However, this release of OH• radicals as a
result of O—O bond homolysis has recently been questioned [32]. It is worth noting
that molecular orbital calculations by Rauk and co-workers [2] have shown that
ONOOH can abstract hydrogen atoms from aliphatic compounds directly. Calculated
results here clearly indicate that decomposition of ONOO− into NO2• and O•− is
unlikely to happen at physiological temperatures, given the high enthalpy of reaction.
The activation barrier for such a process is expected to exceed 240 kJ mol–1. The
decomposition of ONOO− into NO− and singlet O2 has been ruled out [33]. CASSCF
calculations carried out on this reaction confirm that this fragmentation has a rather
high enthalpy, thus this decomposition pathway was not considered in detail. Whether
these decomposition reactions of ONOOH and ONOO− actually occur in a
straightforward manner is yet to be proven. Although potential energy surfaces for
ONOOH has been calculated and extensively mapped out [34–36], information
related to the transition states corresponding to O—O bond breaking in ONOO− and
ONOOH is lacking. Calculations of these enthalpies at high levels of theory can
provide at least some insights into the feasibility of these fragmentation processes.
3.4 Conclusions
The heats of formation, ΔH°f,298, for a series of sixteen biologically relevant RNOS
were calculated using various composite methods. The G3 and G3X(MP2) methods
give the best performance with a mean absolute deviation from experiment of 5.1 and
69
4.9 kJ mol–1 respectively. However, the less impressive performance of the G3X
method compared with the G3X(MP2) method suggests that the apparent good
performance of G3X(MP2) could be fortuitous. The best estimates for the heats of
formation of ONOO– and ONOOH are –108 and –18 kJ mol–1, respectively. The
reaction enthalpies for these reactions involving ONOO– and ONOOH decomposition
were also examined at various levels of theory. The geometry of ONOO– is found to
be poorly described by the QCISD method. As a result, CBS-APNO and CBS-RAD
give rather unsatisfactory thermochemistry related to peroxynitrite.
References
1. Wink, D. A.; Mitchell, J. B. Free Rad. Biol. Med. 1998, 25, 434.
2. Shustov, G. V.; Spinney, R.; Rauk, A. J. Am. Chem. Soc. 2000, 122, 1191.
3. Hodgen, B.; Rayat, S.; Glaser, R. Org. Lett. 2003, 5, 4077.
4. Szabó, C. Toxicol. Lett. 2003, 140-141, 105.
5. Aplincourt, P.; Bohr, F.; Ruiz-Lopez, M. F. J. Mol. Struc. (THEOCHEM)
1998, 426, 95.
6. Dedon, P. C.; Tannenbaum, S. R.; Arch. Biochem. Biophys. 2004, 423, 12.
7. Alvarez, B.; Radi, R.; Amino Acids, 2003, 25, 295.
8. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Rassolov, V.; Pople, J. A. J.
Chem. Phys. 1999, 110, 4703.
9. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Pople, J. A. J. Chem. Phys.
2001, 114, 108.
10. Curtiss, L. A.; Raghavachari, K.; Redfern, P. C.; Pople, J. A. J. Chem. Phys.
1998, 109, 7764.
70
11. Ochterski, J. W.; Petersson, G. A.; Montgomery, Jr., J. A. J. Chem. Phys. 1996,
104, 2598.
12. Montgomery, Jr., J. A.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J.
Chem. Phys. 2000, 112, 6532.
13. Petersson, G. A.; Tensfeldt, T. G.; Montgomery, Jr., J. A. J. Chem. Phys. 1991,
94, 6091.
14. Montgomery, Jr., J. A.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J.
Chem. Phys. 1999, 110, 2822.
15. Montgomery, Jr., J. A.; Ochterski, J. W.; Petersson, G. A. J. Chem. Phys. 1994,
101, 5900.
16. Mayer, P. M.; Parkinson, C. J.; Smith, D. M.; Radom, L. J. Chem. Phys. 1998,
108, 604.
17. Henry, D. J.; Sullivan, M. B.; Radom, L. J. Chem. Phys. 2003, 118, 4849.
18. Nicolaides, A.; Rauk, A.; Glukhovtsev, M. N.; Radom, L. J. Phys. Chem. 1996,
100, 17460.
19. Chase, Jr., M. W. NIST-JANAF Thermochemical Tables; 4th Ed.; American
Chemical Society: Woodbury, NY, 1998.
20. Tsai, H. –H.; Hamilton, T. P.; Tsai, J. –H. M.; van der Woerd, M.; Harrison, J.
G.; Jablonsky, M. J.; Beckman, J. S.; Koppenol, W. H. J. Phys. Chem. 1996,
100, 15087.
21. Woerle, M.; Latel, P.; Kissner, R.; Nesper, R.; Koppenol, W. H. Chem. Res.
Toxicol. 1999, 12, 305.
22. Werner, H.-J.; Knowles, P. MOLPRO 2002, University of Birmingham.
23. Gaussian 98 (Revision A.7), Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.;
Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.;
71
Montgomery Jr., J. A.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J.
M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone,
V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P.
M. W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A.; Gaussian, Inc., Pittsburgh, PA, 1998.
24. Traverst, M. J.; Cowles, D. C.; Ellison, G. B. Chem. Phys. Lett. 1989, 164, 449.
25. Nobes, R. H.; Moncrieff, D.; Wong, M. W.; Radom, L.; Gill, P. M. W.; Pople,
J. A. Chem. Phys. Lett. 1991, 182, 216.
26. Romero, N.; Radi, R.; Linares, E.; Augusto, O.; Detweiler, C. D.; Mason, R.
P.; Denicola, A. J. Biol. Chem. 2003, 278, 44049.
27. Kazazić, S.; Kazazić, S. P.; Klasinc, L.; McGlynn, S. P.; Pryor, W. A. J. Phys.
Org. Chem. 2002, 15, 728.
28. Janoschek, R.; Kalcher, J. Z. Anorg. Allg. Chem. 2002, 628, 2724.
29. Olson, L. P.; Bartberger, M. D.; Houk, K. N. J. Am. Chem. Soc. 2003, 125,
3999.
30. Pfeiffer, S.; Mayer, B.; Janoschek, R. J. Mol. Struct. (THEOCHEM) 2003, 623,
95.
31. Beckman, J. S.; Beckman, T. W.; Chen, J.; Marshall, P. A.; Freeman, B. A.
Proc. Natl. Acad. Sci. USA 1990, 87, 1620.
72
32. Kissner, R.; Nauser, T.; Kurz, C.; Koppenol, W. H. IUBMB Life 2003, 55,
567.
33. Merényi, G.; Lind, J.; Czapski, G.; Goldstein, S. Proc. Natl. Acad. Sci. USA
2000, 97, 8216.
34. Zhu, R. S.; Lin, M. C. J. Chem. Phys. 2003, 119, 10667.
35. Bach, R. D.; Dmitrenko O.; Estévez, C. M. J. Am. Chem. Soc. 2003, 125,
16205.
36. Li, Y.; Francisco, J. S. J. Chem. Phys. 2000, 113, 7976.
73
Chapter Four
Decomposition of Nitrosoperoxycarbonate Anion: Gas
Phase and Solvent Modeled Studies.
4.1 Introduction
Peroxynitrite (ONOO–), formed from the interaction of nitric oxide (NO•) and
superoxide (O2•–) [1], is a potent two electron oxidant responsible for direct and
indirect effects of reactive nitrogen oxide species. ONOO– reacts with several
substrates such as dimethylsulfoxide and deoxyribose with zero order kinetics [2], and
with substrates such as methionine [3] and ascorbate [4] with small second-order rate
constants.
Biologically, ubiquitous CO2 can accelerate oxidation of the above substrates by
ONOO– [5, 6]. The direct reaction of ONOO– with CO2 produces an intermediate –
nitrosoperoxycarbonate, ONOOCO2– [7] – which undergoes homolytic scission of the
O—O bond to yield NO2• and CO3
•– radicals [8-11].
ONOO– + CO2 → ONO−OCO2– → NO2
• + CO3•–
Previous theoretical calculations have shown that the O—O bond in ONOOCO2– is
exceptionally weak [12] and this is substantiated by the derivation of the very high
overall rate constant that governs this process (k = 1.9 × 109 s–1) [11]. These radicals
74
can diffuse out and initiate other free radical reactions, playing a role in various
pathophysiological processes [13]. The CO3•– radical is relatively less reactive than
•OH, but being more stable, it can diffuse further from the site origin and propagate
oxidative damage. NO2• is able to initiate free radical reactions, for instance, hydrogen
abstraction from fatty acids, tyrosine residues or tocopherol [6, 14].
Currently, due to the transient nature of the reactive species discussed, it is difficult to
study the factors that govern the formation of the free radicals NO2• and CO3
•– that are
reported to cause so much harm. To this end, a computational study of the reaction
shown above, in the gas phase and in solvent, is attempted and presented.
Modeling of the solvent environment can be carried out in using implicit solvent
models (ISM), explicit solvent models (ESM) and hybrid solvent models (HSM). ISM,
or continuum models of solvent, focuses on dielectric models of electrostatic effects.
The solvent is modeled as a continuous, homogeneous and isotropic medium and is
characterized by a scalar value – the dielectric constant ε [15, 16]. ESM involves the
addition of discrete solvent molecules around the solute to constitute the first
solvation shell, and subsequent calculations for this system carried out in the gas
phase. HSM used in this work involves the addition of solvent molecules to constitute
the first solvation shell of the solute as in ESM, followed by the treatment of this
solvent-bound complex using ISM, similar to the approach adopted by Kiruba et al in
the study of tautomeric equilibria of 3-hydroxypyridine derivatives [17]. Calculations
in the gas phase and in solvent using various approaches would provide valuable
information regarding solvent effect on the reaction profile.
75
The detection and monitoring of reactive species related to ONOO– reaction with CO2
is commonly carried out using UV-visible spectroscopy. For instance, it was proposed
by Koppenol et al that the adduct formed by the reaction of ONOO– with CO2 absorbs
at 640 nm [9]. In an attempt to provide a better understanding of this observation, the
absorption spectra of the reactive species ONOO–, ONOOCO2– and CO3
•– were also
studied using various computational methods.
4.2 Computational Methods
The hybrid B3LYP functional [18], a three-parameter combination of Becke’s 1988
exchange functional [19], the Lee-Yang-Parr functional for non-local correlation [20]
and the Vosko-Wilk-Nusair functional III for local correlation [21] was employed for
DFT calculations. Full geometry optimizations of all species studied were carried out
at the B3LYP/6-31+G(d) level. Frequency calculations were carried out for the
stationary points obtained, to determine whether they were indeed local minima with
all real harmonic frequencies, or first order saddle points, with one imaginary
frequency. Additional higher level energy calculations were carried out at the
B3LYP/6-311+G(2df,p) level, based on B3LYP/6-31+G(d) geometries. Selected gas-
phase benchmark calculations were performed at the G3 [22], CBS-QB3 [23] and
CCSD(T)/aug-cc-pVTZ//CCSD/aug-cc-pVDZ levels of theory.
The continuum solvation models employed were the Onsager self consistent reaction
field model (SCRF=Dipole) [24-26] and the integral equation formalism for the
polarizable continuum model or IEFPCM [27-29]. The dielectric constant ε, used in
76
these models is 78.39, corresponding to that for water. For the IEFPCM methods, the
united atom topological model (UA0) method was used to generate atomic radii for
construction of the molecular cavity. Calculations using explicit and hybrid solvation
models used four discrete water molecules to form the first solvation shell.
For calculation of the gas-phase transition energies (λexcitation) of ONOO–, ONOOCO2–
and CO3•–, Zerner’s intermediate neglect of differential overlap (ZINDO/S) [30],
single-excitation configuration interaction (CIS) [31] and time-dependent density
functional theory (TD-DFT) [32, 33] methods were employed, based on gas-phase
geometries obtained at the B3LYP/6-31+G(d) level.
Closed shell species were treated using the restricted Hartree-Fock (RHF) formalism
and open shell species using the unrestricted Hartree-Fock (UHF) formalism, except
for CCSD calculations, in which the ROHF reference wave functions are used for
open-shell systems instead. Singlet diradical species were treated using UHF, with the
HOMO and LUMO mixed so as to destroy α-β and spatial symmetries. Spin
correction for open shell species were treated using the approximate spin correction
procedure proposed by Yamaguchi et al [34, 35]. Unless otherwise noted, energies
reported are calculated at the B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d) + ZPE +
spin correction level.
All calculations were carried out using the GAUSSIAN 03 [36] and MOLPRO 2002
[37] suite of programs.
77
4.3 Results and Discussion
4.3.1 Gas-phase reaction profile of ONOO– reaction with CO2
The following detailed reaction scheme and labeling of the various species involved is
proposed. Cis- and trans-peroxynitrite (1 and 2, respectively) undergo reaction with
CO2 (3) to form cis- and trans-nitrosoperoxycarbonate (4 and 5). These adducts
undergo O—O bond homolysis, via transition states TS1 and TS2, respectively, to
form nitrogen dioxide (6) and carbonate radical anion (7) (See Scheme 4.1).
Scheme 4.1. Proposed reaction scheme and labeling of various species involved in the reaction ONOO– + CO2 → ONO−OCO2
– → NO2• + CO3
•–.
Structures of the local minima 1 to 7 and transition states TS1 and TS2, calculated at
the B3LYP/6-31+G(d) level, are presented in Figure 4.1. Relative gas-phase energies
78
of these species (including ZPE and spin correction) are presented in Table 4.1, with
G3, CBS-QB3 and CCSD(T)/aug-cc-pVTZ//CCSD/aug-cc-pVDZ energies included
for comparison. ONOO– and ONOOCO2– in the cis conformation is preferred, and
they are 12.3 and 5.1 kJ mol–1 lower in energy than their trans counterparts,
respectively. Preference of the ONOO– molecule to adopt the cis conformation can be
explained by a favorable π*–π* interaction formed from the overlap of O1=N1 and
O2—O3 π* orbitals, as seen in the highest occupied molecular orbital (HOMO) of
cis-ONOO– (See Figure 4.2). The CO2 molecule forms an adduct with ONOO– above
the O–N–O–O plane. The gas-phase dihedral angles N1–O2–O3–C1 for 4 and 5 are –
89.7° and –84.4°, respectively.
79
N1 O2
O1O3
O3
O1N1
O2
C1O1 O2
O2
O3
N1
O4
C1
O1
O5
O3O2
O4C1
N1
O5
O1
O1
N1
O2
O1
O3
C1
O2
O1
N1
O3O4
O2C1
O5
O1
N1
O2
O3
C1
O5
O4
1.222
(1.214)
[1.201]
1.367
(1.364)
[1.391]1.390
(1.402)
[1.427] 1.237
(1.221)
[1.196]
1.339
(1.350)
[1.412]1.410
(1.421)
[1.446]
1.169
(1.169)
[1.169]
1.192
(1.181)
[1.185]
1.381
(1.412)
[1.397]
1.440
(1.442)
[1.449]
1.571
(1.488)
[1.503]
1.223
(1.231)
[1.230]
1.227
(1.239)
[1.242]
1.439
(1.434)
[1.439]
1.383
(1.428)
[1.423]
1.200
(1.183)
[1.185]
1.559
(1.485)
[1.495]
1.225
(1.232)
[1.230]1.228
(1.239)
[1.245]
1.202
(1.201)
[1.202]
1.278
(1.275)
[1.278]
1.217
(1.203)
[1.188]
1.251
(1.240)
[1.257]
1.903
(1.858)
[1.789]
1.363
(1.371)
[1.378]
1.230
(1.239)
[1.247]
1.274
(1.266)
[1.278]
1.235
(1.227)
[1.218]
1.279
(1.278)
[1.295]
2.012
(2.000)
[1.964]
1.319
(1.323)
[1.336]
1.230
(1.233)
[1.236]
1.294
(1.286)
[1.284]
1.169
(1.169)
[1.169]
1.202
(1.201)
[1.202]
1.278
(1.275)
[1.278]
1.278
(1.275)
[1.278]
1 23
4 5
6 7
TS1 TS2
Figure 4.1. Structures of local minima 1-7, TS1 and TS2, calculated at the B3LYP/6-31+G(d) level. Distances are given in Ångstroms. Bond lengths in regular notation, parentheses, and square brackets correspond to gas-phase, IEFPCM solvation modeled, and Onsager’s SCRF solvation modeled values, respectively.
80
Figure 4.2. Highest occupied molecular orbital (HOMO) of cis-ONOO– calculated at the B3LYP/6-31+G(d) level, showing extent of π*–π* interaction.
The reactions 1 + 3 → 4 and 2 + 3 → 5 are exothermic by 61.7 and 56.6 kJ mol–1,
respectively, at the B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d) level, an
underestimation of the binding energy of CO2 to ONOO–, when compared to
benchmark results. Other DFT methods reproduce similar trends in the
underestimation of the binding energy, for instance, the density functionals TPSS and
VSXC with the same basis sets employed yielded binding energies of 62.5 and 59.0
kJ mol–1, respectively. The calculated overall enthalpy at 0 K for the reactions 1 + 3
→ 6 + 7 and 2 + 3 → 6 + 7 generally do not agree well, even among benchmark
calculations. Spin contamination in the CO3•− radical could be the cause of such
discrepancy in the G3 method, since the energies presented are not spin-corrected.
81
The overall enthalpy is expected to be close to the CBS-QB3 and CCSD(T) value, i.e.
−47.3 and −35.4 kJ mol–1, respectively, of which B3LYP does not reproduce too well.
Experimentally, ONOOCO2– is not accumulated during the reaction of ONOO– with
CO2, and cannot be detected spectrophotochemically [1, 7, 38, 39]. Also, attempts to
chemically trap or scavenge ONOOCO2– have not been successful [40]. Calculated
gas-phase O—O homolytic dissociation barriers for cis- and trans-
nitrosoperoxycarbonate are low, at 17.7 and 35.1 kJ mol–1, respectively. The low
barrier for 4 → TS1 shows the relative ease of O—O homolytic bond scission. As the
O—O homolytic dissociation of ONOOCO2– is expected to involve a curve crossing
from the singlet potential energy hypersurface (PES) to the triplet PES of the two
radicals NO2• and CO3
•–, the transition states TS1 and TS2 are expected to be singlet
and diradicaloid in nature, a result of contamination from the triplet state wave
function. The diradical wave function is also expected to be more stable than the
singlet and triplet wave functions. Expectation values of the spin-squared operator
⟩⟨2S for such singlet diradicals are between that of a singlet (i.e. ⟩⟨
2S = 0) and a
triplet (i.e. ⟩⟨2S = 2). A ⟩⟨
2S value of unity indicates 50:50 singlet and triplet, and
the singlet and triplet states are nearly degenerate. The ⟩⟨2S values for TS1 and TS2
are 0.27 and 0.54, and their corresponding spin-corrections to the energy are 11.5 and
16.3 kJ mol–1, respectively.
In our various attempts to find the O2NOCO2– species described by many authors as a
rearrangement product of ONOOCO2–, no local minima were found that corresponded
to the nitrogen in NO2• forming a proper covalent bond with oxygen in CO3
•−. This is
substantiated by previous theoretical work that concluded that O2NOCO2– is not a
82
local energy minimum in the PES [41]. The overall process is expected to proceed in
the following scheme: 1 + 3 → 4 → TS1 → 6 + 7.
Species B3LYPa G3 CBS-QB3 CCSD(T)b
1 + 3 0.0 0.0 0.0 0.0 2 + 3 12.3 13.3 14.0 16.7 4 −61.7 −73.7 −78.2 −77.7 5 −56.6 −66.9 −73.0 −74.9
6 + 7 −68.8 −19.5 −47.3 −35.4
TS1 −44.0 (17.1) TS2 −21.5 (35.1)
Table 4.1. Zero point corrected relative energies of 1-7, TS1 and TS2, in kJ mol–1. Values in parentheses indicate intrinsic barriers for 4 → TS1 and 5 → TS2. aB3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d) + ZPE + spin correction. bCCSD(T)/aug-cc-pVTZ//CCSD/aug-cc-pVDZ + ZPE.
4.3.2 Implicit Solvation Model (ISM) Results
The gas-phase structures 1-7, TS1 and TS2 were re-optimized using the Onsager
SCRF model and Tomasi’s IEFPCM solvation models for implicit solvation, to
examine the bulk solvent effects. Calculated relative energies of the species involved
are shown in Table 4.2 and geometric parameters of the local minima and transition
states using ISM are shown in Figure 4.1.
Species ISM (Onsager) ISM (IEFPCM)
1 + 3 0.0 0.0 2 + 3 6.6 16.4 4 −52.9 −17.2 5 −49.6 −9.5
6 + 7 −12.5 −54.4
TS1 −51.5 (1.4) 0.7 (17.9) TS2 2.2 (51.8) 42.3 (51.8)
Table 4.2. B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d) + ZPE and spin correction relative energies of 1-7, TS1 and TS2, in kJ mol–1, using implicit solvation models. Values in parentheses indicate intrinsic barriers for 4 → TS1 and 5 → TS2.
83
Compared to the gas phase and ISM (IEFPCM) values, the cis- and trans-ONOO–
energy difference for ISM (Onsager) appears to be understated, at 6.6 kJ mol–1. The
binding energy of cis- and trans-ONOO– to CO2 differs greatly across both methods
as well, from −52.9 (1 + 3 → 4) and −56.2 kJ mol–1 (2 + 3 → 5) in ISM (Onsager) to
−17.2 (1 + 3 → 4) and −25.9 kJ mol–1 (2 + 3 → 5) in ISM (IEFPCM). The overall
energy change from 1 + 3 → 6 + 7 and 2 + 3 → 6 + 7 seems to differ greatly across
both methods as well. However, both ISMs employed predict low reaction barriers for
4 → TS1, at 1.4 kJ mol–1 for ISM (Onsager) and 17.9 kJ mol–1 for ISM (IEFPCM),
and a higher barrier for 5 → TS2, at 51.8 kJ mol–1 for both methods.
As seen from Figure 4.1, the incorporation of ISM did not drastically change the
structures of 1-7, TS1 and TS2. Particularly, in the small molecules 3, 6 and 7,
geometric parameters remain somewhat unchanged. The C=O bond lengths in CO2
show no change at 1.169 Å in the gas phase, ISM (IEFPCM) and ISM (Onsager)
models. For NO2• and CO3
•–, N—O and C—O bond lengths are only slightly shorter
at 1.201 and 1.275 Å, respectively, compared to 1.202 and 1.278 Å for gas-phase and
ISM (Onsager) modeled geometries. In cis-ONOO–, the N1=O1 bond is shortened
slightly, from 1.222 Å in the gas phase, to 1.214 Å for ISM (IEFPCM) and 1.201 Å
for ISM (Onsager). The N1—O2 bond length is lengthened, from 1.367 Å in the gas
phase to 1.364 Å for ISM (IEFPCM) and 1.391 Å for ISM (Onsager). The O2—O3
bond is also lengthened, from 1.390 Å in the gas phase to 1.402 Å for ISM (IEFPCM)
and 1.427 Å for ISM (Onsager). Similar trends are seen for trans-ONOO–; the
terminal N=O bond is shortened, the central N—O and the terminal O—O bonds are
lengthened in solution. Again, there are no significant changes in bond lengths for cis-
84
and trans-ONOOCO2–. However it should be noted that C1—O3 bond lengths show
noticeable shortening from gas phase to solution. In cis-ONOOCO2–, the C1—O3
bond lengths shortened from 1.571 Å to 1.488 and 1.503 Å for ISM (IEFPCM) and
ISM (Onsager), respectively, and in trans-ONOOCO2–, they shortened from 1.559 Å
to 1.485 and 1.495 Å.
The O2—O3 bond lengths in cis- and trans-ONOOCO2– are similar in length, at about
1.44 Å in both gas phase and in solvent. It is also observed that both ISMs predict a
lowering of the O2—O3 distance in TS1 and TS2, more so for TS1, where the O2—
O3 distance is shortened from 1.903 Å in the gas phase to 1.789 Å for ISM (Onsager).
However, the O—O bond dissociation energies (BDE) in ONO–OCO2– differ greatly
in gas-phase and solvent modeled calculations. O2—O3 bond lengths are seen here as
not being reflective of the actual BDE.
The lack of consistency in the energy results yielded from both ISMs employed
suggests that the use of continuum methods alone is insufficient to adequately
describe the solvent effects on the energetics of this system. There is a need, therefore,
to use methods that can account for more than just the nonspecific solute-solvent
interactions as considered in ISMs. A closer look at solvent effects of the first
solvation shell is described in the following sections.
4.3.3 Explicit Solvation Model (ESM) Results
Discrete solvation of the solutes (ESM) was carried out as described in the following
reaction scheme:
85
(H2O)3·ONOO– + (H2O)·CO2 → (H2O)4·ONOOCO2– → (H2O)·NO2
• + (H2O)3·CO3•–
Four of the solvent molecules are treated as discrete entities, used to describe the
microscopic properties of the solvent at the first solvation shell. The structures of
these clusters (local minima and transition states) 1a-7a, TS1a and TS2a, calculated
using ESM are shown in Figures 4.3 and 4.4. Relative energies of these solutes
B3LYP/6-311+G(2df,p)//B3LYP/6-31G(d) + ZPE + Spin correction level are shown
in Table 4.3.
Species ESM (4H2O) 1a + 3a 0.0 2a + 3a 14.9 4a −42.2 5a −41.4
6a + 7a −32.7
TS1a −22.9 (19.3) TS2a 9.3 (50.7)
Table 4.3. B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d) + ZPE + Spin correction relative energies of 1a-7a, TS1a and TS2a, in kJ mol–1, using the explicit solvation model (ESM). Values in parentheses indicate intrinsic barriers for 4a → TS1a and 5a → TS2a.
86
O6
O4
O3
O2
O1
N1
O5
O2
O3
C1O1
N1
O2O1
O3
O4
O3O2
C1
O5O1
O6
O3
N1
O2
O4
O1
O5
O6
2.771
(2.771)
2.850
(2.649)
1a 2a
3a 6a7a
1.212
(1.203)
1.360
(1.376)
1.405
(1.413)
1.214
(1.207)
1.359
(1.374)
1.426
(1.429)
1.168
(1.170)
1.171
(1.170)
1.201
(1.203)
1.201
(1.203)
1.290
(1.293)
1.280
(1.275)
1.262
(1.265)
Figure 4.3. Structures of local minima 1a, 2a, 3a, 6a and 7a, calculated at the B3LYP/6-31+G(d) level. Distances are given in Ångstroms. Bond lengths in regular notation and parentheses correspond to gas-phase (ESM), and Onsager’s SCRF solvation modeled (HSM) values, respectively. Dotted lines show hydrogen bond interactions (1a, 2a, 7a) and dipole-dipole interactions (3a, 6a) between solute and solvent, which have distances less than the sum of van der Waals’ radii for the two atoms concerned.
87
O8
O9
O2O3
O4
O7
C1
N1
O5
O1
O6
O9
O8
O2O3
O7
O1 N1
C1
O5
O4
O6
O2
N1
O7
O3
O9
O1
C1
O5
O4
O8
O6
O9
O7
O1 N1
O2
O5
O6
O3
C1
O4
O8
1.869
(1.849)
1.964
(1.953)
4a 5a
TS1a TS2a
1.181
(1.174)
1.416
(1.441)1.437
(1.436)
1.514
(1.479)
1.230
(1.234)
1.231
(1.240)
1.188
(1.178)
1.414
(1.448)
1.434
(1.431)
1.473
(1.453)
1.237
(1.236)
1.238
(1.248)
1.204
(1.199)
1.248
(1.239)1.370
(1.371)
1.238
(1.242)
1.261
(1.265)
1.228
(1.226)
1.268
(1.270)
1.328
(1.330)
1.239
(1.241)
1.277
(1.275)
Figure 4.4. Structures of local minima 4a and 5a, and transition states TS1a and TS2a, calculated at the B3LYP/6-31+G(d) level. Distances are given in Ångstroms. Bond lengths in regular notation and parentheses correspond to gas-phase (ESM), and Onsager’s SCRF solvation modeled (HSM) values, respectively. Dotted lines show hydrogen bonding interactions between solute and solvent, which have distances less than the sum of van der Waals’ radii for the two atoms concerned.
The preferred sites for hydrogen bonding in 1a and 2a are on O3 (See Figure 4.3). It
is interesting to note, that for 1a, all three water molecules exclusively bind to O3,
compared to two to O3 and one to O2 in 2a. Solvation studies of ONOO– using the
PBE functional predicted a similar outcome with two discrete solvent molecules [42].
NBO charge analysis of 1 and 2 at the B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d)
level shows that the charge on O3 is the most negative in –0.592 for 1, and –0.603 for
88
2, hence O3 would most likely be the preferred site for hydrogen bond formation
between water molecules and the solutes 1 and 2. Attempts to optimize 6a and 7a
with the water molecule forming a hydrogen bond with oxygen atoms in CO2 and
NO2• as starting structures resulted in the local minimum structure of the dipole-dipole
complex shown in Figure 4.3. The C1–O3 and N1–O3 distances are still less than the
sum of the van der Waals’ radii of the atoms involved, which are 3.22 and 3.07 Å,
respectively. In 7a, water molecules are found to bind to O2 and O3 via one hydrogen
bond per water molecule. The approximate molecular symmetry of 7a is D3h, while
the actual symmetry is C1. Attempts to optimize the structure at D3h symmetry
resulted in a number of small imaginary frequencies corresponding to rotations of
water molecules along the plane defined by CO3•–. The tendency for water to form
one hydrogen bond interaction instead of two is also shown in the results of
Armstrong et al [43], where the optimization of (H2O)·CO3•– resulted in a CS structure
instead of a C2v structure. The four water molecules in 4a, and TS1a show preference
of bonding to the CO3 moiety, whereas for 5a and TS2a, one water molecule is shown
to bind to the NO2 moiety and the other three to the CO3 moiety. Calculated bond
lengths of the solutes 1a-7a, TS1a and TS2a using ESM appear to be consistent with
those obtained for ISM (IEFPCM) and ISM (Onsager) (See Figure 4.1) in the
previous section.
The energy difference between the solvated forms of the two cis- and trans-ONOO–
conformers (1a and 2a) in the ESM is 14.9 kJ mol–1, similar to that calculated in the
gas phase. The binding energy of cis- and trans-ONOO– to CO2 is less exothermic
than gas-phase binding energies, at –42.2 and –56.3 kJ mol–1, for 1a + 3a → 4a and
2a + 3a → 5a, respectively. The overall reaction energy is less exothermic, at –32.7
89
and –47.6 kJ mol–1 for the reactions 1a + 3a → 6a + 7a and 2a + 3a → 6a + 7a,
respectively, compared to the B3LYP gas-phase results of –68.8 and –81.1 kJ mol–1.
4.3.4 Hybrid Solvation Model (HSM) Results
The structures obtained in ESM were re-optimized using the Onsager SCRF, using a
dielectric constant ε = 78.39. This method provides both microscopic (discrete water
molecules) and macroscopic representations of the solvent in theoretical treatments to
the solutes. Relative energies of the species 1a-7a, TS1a and TS2a are shown in
Table 4.4, and selected geometric parameters are shown in Figures 4.3 and 4.4.
Species HSM (Onsager, 4H2O) 1a + 3a 0.0 2a + 3a 19.8 4a −32.4 5a −32.0
6a + 7a −25.4
TS1a −18.0 (14.4) TS2a 32.7 (64.7)
Table 4.4. B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d) + ZPE + Spin correction relative energies of 1a-7a, TS1a and TS2a, in kJ mol–1, using the hybrid solvation model (HSM). Values in parentheses indicate intrinsic barriers for 4a → TS1a and 5a → TS2a.
Molecular geometries of 1a-7a, TS1a and TS2a for HSM do not differ much from
those obtained in ESM, which are in turn consistent with those obtained using ISM
(IEFPCM) and ISM (Onsager), as shown in Figures 4.3 and 4.4, thus it is noted, that
these geometric parameters are not changed significantly with increasing
sophistication of solvation models.
90
The energy gap between cis- and trans-ONOO– in HSM is calculated to be 19.8 kJ
mol–1, which is slightly greater than those predicted in gas-phase, ISM and ESM
calculations. Binding energies of ONOO– to CO2 are –32.4 and –51.8 kJ mol–1 for 1a
+ 3a → 4a and 2a + 3a → 4a, respectively, which are slightly less exothermic than
those for ESM (–42.2 and –56.3 kJ mol–1). The overall reaction energy for 1a + 3a →
6a + 7a and 2a + 3a → 6a + 7a is –25.4 and –45.2 kJ mol–1, respectively, again
slightly less exothermic than those for ESM (–32.7 and –47.6 kJ mol–1). Incorporation
of a solvent reaction field to the solute-solvent clusters in ESM is observed to lower
reaction energies. Another interesting observation is that the energy difference
between the cis and trans conformers 4a and 5a is very small, at 0.4 kJ mol–1 in HSM.
The intrinsic activation barrier 4a → TS1a has decreased from 19.3 kJ mol–1 (ESM)
to 14.4 kJ mol–1 (HSM) upon the incorporation of a solvent reaction field, while the
barrier for 5a → TS1a has increased from 50.7 kJ mol–1 to 64.7 kJ mol–1 (HSM).
So far, continuum approaches and first solvation shell approaches have been
employed to model the reaction profile in an aqueous solvent. It is noted, however,
that there are still problems with such approaches. Applicability of continuum
approaches is also limited by the specific interactions of the solvent molecules with
the solute, effects of which dominate in the first solvation shell. In first solvation shell
approaches (e.g. ESM and HSM), it is not clear how the number and orientation of
non-bulk water molecules should be determined. There are also a very large number
of states which correspond to local minima, and the averaging of these states should
be considered as representative of the solution described [15]. QM/MM approaches
were adopted by González Lebrero et al in the study of solvated cis-ONOO– [42], and
these showed similar results to structures obtained here using ESM and HSM. The use
91
of QM/MM is attractive, but will still be computationally more expensive than ISM
approaches. Nevertheless, there remains a need for quantitatively accurate solvation
models without incurring significant computational cost that can provide a definitive
description of the system studied. The HSM approach used is perhaps the closest that
has been achieved here.
4.3.5 UV-visible spectra of ONOO–, ONOOCO2
– and CO3
•–
UV-visible absorption bands of cis-ONOO– (1), trans-ONOO– (2), cis-ONOOCO2–
(4), trans-ONOOCO2– (5), and CO3
•– (7) were predicted using ZINDO/S, CIS/6-
311+G(2df,p) and TD-B3LYP/6-311+G(2df,p) levels using geometries obtained at the
B3LYP/6-31+G(d) level. Table 4.5 shows the bands obtained and their corresponding
oscillator strengths.
ZINDO/S CISa TD-B3LYPb IEFPCM-TD-B3LYPc
Species Symd λexc/nm f λexc/nm f λexc/nm f λexc/nm f
1 A′ 445.0 0.098 209.3 0.128 312.6 0.092 304.1 0.107 2 A′ 420.7 0.117 213.1 0.168 366.1 0.095 349.9 0.110 4 A 326.9 0.000 152.0 0.001 373.3 0.001 351.6 0.001 5 A 325.6 0.004 152.1 0.001 393.5 0.002 374.7 0.002 7 E′ 165.4 0.157 218.8 0.039 603.1 0.025 620.3 0.030
Table 4.5. Predicted UV-vis absorption bands of cis-ONOO– (1), trans-ONOO– (2), cis-ONOOCO2
– (4), trans-ONOOCO2
– (5), and CO3•– (7) using various methods, based geometry obtained at B3LYP/6-
31+G(d) level. aCIS/6-311+G(2df,p). bTD-B3LYP/6-311+G(2df,p). cTD-B3LYP/6-311+G(2df,p) in IEFPCM SCRF, solvent dielectric=78.39. dTransition symmetry label.
Absorption wavelengths of cis- and trans-ONOO– were previously calculated by
Krauss to be 324.3 and 368.5 nm, respectively, using MCSCF methods [44]. This
λexcitation is calculated to be blue-shifted in a solvent reaction field, to 310.4 and 338.8
92
nm. The experimental absorption spectra of cis- and trans-ONOOK isolated in solid
Ar and were obtained by Lo et al, with absorption wavelengths of 325 and 375 nm
[45]. Gas-phase TD-DFT calculated values (312.6 and 366.1 nm) are in good accord
with the experimental and MCSCF-calculated values, while the ZINDO/S and CIS
methods yield significant deviations (≥ 100 nm). This relatively good agreement of
TD-DFT predicted values provides some confidence in the applicability of the TD-
DFT method to the other species investigated.
Electronic transitions in cis- and trans-ONOOCO2– (4 and 5, respectively) occur at
373.3 and 393.5 nm, respectively, in the gas phase. Higher-level calculations were
also performed using Nakatsuji’s symmetry-adapted-cluster configuration-interaction
(SAC-CI) methods [46-48] incorporated into the GAUSSIAN 03 code. Absorption
bands of cis- and trans-ONOOCO2– at the SAC-CI/6-31+G(d)//B3LYP/6-31G(d)
level are 372.9 (f = 0.0025) and 392.4 (f = 0.0021) nm, respectively. The almost
perfect agreement of the TD-DFT results with SAC-CI results lends further
confidence that the TD-DFT method was a good choice for elucidating the absorption
bands of the cis- and trans-ONOOCO2– system. As with cis- and trans-ONOO–, a
blue-shift is expected in a polar solvent reaction field, to 351.6 and 374.7 nm.
However, the oscillator strength f for these transitions is low, between 0.001 and
0.002. Additional TD-DFT calculations were carried out for cis-ONOOCO2 – with
two explicit water molecules, incorporating bulk solvent effects using IEFPCM, at the
B3LYP/aug-cc-pVTZ level. At this level of theory, no electronic transitions above
450 nm were obtained. This could partially explain why ONOOCO2– is not detected
spectrophotometrically, as electronic transitions for both cis- and trans-ONOOCO2–
do not occur within the UV and visible regions in the electromagnetic spectrum.
93
Although some authors claim that ONOOCO2– was indeed spectrophotometrically
observed as an intermediate in the reaction of ONOO– with CO2 [9], it has also been
reported that the observed peak is due to the effect from stray light [49].
The observed band at 640 nm in the reaction of ONOO– with CO2, which decreases in
intensity as the reaction proceeds, is probably not due to the adduct ONOOCO2–, but
possibly CO3•–. The calculated absorption band for CO3
•– using TD-DFT is at 603.1
nm. However, upon incorporating solvent effects using IEFPCM, this band is
observed to be red-shifted to 620.3 nm. Upon calculation of the absorption spectrum
for (H2O)·CO3•– using TD-B3LYP/6-311+G(2df,p)//B3LYP/6-31+G(d), it was found
that the band was split to the two A′ states, 624.3 (f = 0.0152) and 593.3 (f = 0.0303)
nm. When IEFPCM SCRF was incorporated (i.e. IEFPCM-TD-B3LYP/6-
311+G(2df,p)//B3LYP/6-31+G(d)), these bands were further red-shifted to 650.6 (f =
0.0177) and 602.9 (f = 0.0368) nm, respectively. The good correspondence of the
calculated 650.6 nm band with the broad 640 nm band in the spectrum obtained by
Meli et al lends weight to the claim that what was spectrophotometrically observed
100 ms after mixing ONOO– with CO2 was CO3•– instead of ONOOCO2
– as
previously concluded [9]. However, the observed red-shift of the 603.1 nm absorption
band to 650.6 nm upon solvation is unusual, as a previous study by Armstrong et al
reported this band to be blue-shifted and split by explicit solvation from 608.2 nm (E′)
to 596.0 (A′) and 569.3 nm (A′) [43] using IEFPCM-TD-B3LYP/6-311+G(3df,2p).
In an attempt to clarify if the CO3•– absorption band at ~600 nm is indeed red-shifted
or blue-shifted upon solvation by an aqueous solvent, the spectra of CO3•–,
(H2O)·CO3•–, and (H2O)2·CO3
•– were calculated using IEFPCM-TD-B3LYP/aug-cc-
94
pVTZ based on IEFPCM-B3LYP/aug-cc-pVTZ geometries. The results are shown in
Table 4.6. It is observed that the addition of coordinated water molecules to CO3•–
actually blue-shifts the absorption band from 605.5 nm (no explicit H2O) to 573.6 nm
(two explicit H2O molecules). These results are now in accord with the results found
by Armstrong et al.
IEFPCM-TD-B3LYP/aug-cc-pVTZ CO3
•– Explicit H2O molecules (n)
0 1 2
λexcitation (1) / nm 605.5 (E′) 605.2 (A′) 573.6 (A′) f (osc. strength) 0.0292 0.0228 0.0182 λexcitation (2) / nm 605.5 (E′) 572.9 (A′) 551.6 (A′) f (osc. strength) 0.0292 0.0359 0.0374 Table 4.6. Calculated UV-Visible excitation wavelengths λexcitation, and corresponding oscillator strength values, of CO3
•– with 0, 1 and 2 explicit H2O molecules, and incorporating IEFPCM for modeling bulk effects of an aqueous solvent, using the TD-B3LYP method and the aug-cc-pVTZ basis set.
A probable explanation can now be provided for the observed phenomena by Meli et
al, in the light of these calculated TD-DFT results. The observed shift of the 640 nm
absorption band at 100 ms after mixing ONOO– and CO2 to a broad band centered at
620 nm at 170 ms is probably due to the combined effects of increasing explicit
solvation of CO3•– and the depletion of CO3
•– due to its reaction with other species in
the buffered solution. This significant solvent effect leading to the blue-shift in the
CO3•– absorption spectrum provides theoretical evidence to warrant some caution in
the interpretation of spectrophotochemical data obtained from the study of this
intriguing reaction.
95
4.4 Concluding Remarks
The potential energy surface of the reaction between ONOO– and CO2 was explored
using DFT calculations, both in the gas-phase and using various solvent models. The
overall reaction is expected to proceed via the lower energy path, that is, cis-ONOO–
+ CO2 → cis-ONOOCO2– → TS1 → NO2
• + CO3•–. The calculated gas-phase overall
reaction energy is exothermic by 68.8 kJ mol–1. Upon the incorporation of an aqueous
solvent, the reaction energy becomes less exothermic. The barrier for O—O
homolytic bond scission is expected to be about 17 kJ mol–1 both in the gas phase and
in an aqueous solvent. Electronic transition energies were calculated for cis-ONOO–,
trans-ONOO–, cis-ONOOCO2–, trans-ONOOCO2
–, and CO3•–, using ZINDO/S, CIS
and TD-DFT methods, of which TD-DFT results agreed best with experimental
absorption wavelengths and benchmark SAC-CI calculations. The TD-DFT results
suggest that the observation of a 640 nm band in the reaction of ONOO– with CO2 is
not due to ONOOCO2– but CO3
•–, and that the solvation effect of an aqueous solvent
is responsible for the blue shift in the absorption spectrum of CO3•–.
References
1. Squadrito, G. L.; Pryor, W. A. Free Radical Biol. Med. 1998, 25, 392.
2. Beckman, J. S.; Beckman, T. W.; Chen, J.; Marshall, P. A.; Freeman, B. A.
Proc. Natl. Acad. Sci. U. S. A. 1990, 87, 1620.
3. Pryor, W. A.; Jin, X.; Squadrito, G. L. Proc. Natl. Acad. Sci. U. S. A. 1994, 91,
11173.
96
4. Squadrito, G. L.; Jin, X.; Pryor, W. A. Arch. Biochem. Biophys. 1995, 322, 53.
5. Uppu, R. M.; Squadrito, G. L.; Pryor, W. A. Arch. Biochem. Biophys. 1996,
327, 335.
6. Veselá, A.; Wilhelm, J. Physiol. Res. 2002, 51, 335.
7. Lymar, S. V.; Hurst, J. K. J. Am. Chem. Soc. 1995, 117, 8867.
8. Goldstein, S.; Czapski, G. J. Am. Chem. Soc. 1998, 120, 3458.
9. Meli, R.; Nauser, T.; Koppenol, W. H. Helv. Chim. Acta 1999, 82, 722.
10. Meli, R.; Nauser, T.; Latal, P.; Koppenol, W. H. J. Biol. Inorg. Chem. 2002, 7,
31.
11. Squadrito, G. L.; Pryor, W. A. Chem. Res. Toxicol. 2002, 15, 885.
12. Houk, K. N.; Condroski, K. R.; Pryor, W. A. J. Am. Chem. Soc. 1996, 118,
13002.
13. Augusto, O.; Bonini, M. G.; Amanso, A. M.; Linares, E.; Santos, C. X. C.; de
Menezes, S. L. Free Radical Biol. Med. 2002, 32, 841.
14. Santos, C. X. C.; Bonini, M. G.; Augusto, O. Arch. Biochem. Biophys. 2000,
377, 146.
15. Cramer, C. J.; Truhlar, D. G. Chem. Rev. (Washington, DC, U. S.) 1999, 99,
2161.
16. Tomasi, J.; Persico, M. Chem. Rev. (Washington, DC, U. S.) 1994, 94, 2027.
17. Kiruba, G. S. M.; Wong, M. W. J. Org. Chem. 2003, 68, 2874.
18. Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
19. Becke, A. D. Phys. Rev. A 1988, 38, 3098.
20. Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785.
21. Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200.
97
22. Curtiss, L. A.; Raghavachari, K.; Redfern, P. C.; Rassolov, V.; Pople, J. A. J.
Chem. Phys. 1998, 109, 7764.
23. Montgomery Jr., J. A.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J.
Chem. Phys. 1999, 110, 2822.
24. Wong, M. W.; Wiberg, K. B.; Frisch, M. J. Chem. Phys. 1991, 95, 8991.
25. Wong, M. W.; Wiberg, K. B.; Frisch, M. J. J. Am. Chem. Soc. 1991, 113, 4776.
26. Wong, M. W.; Wiberg, K. B.; Frisch, M. J. J. Am. Chem. Soc. 1992, 114, 1645.
27. Cancès, E.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107, 3032.
28. Cossi, M.; Barone, V.; Mennucci, B.; Tomasi, J. Chem. Phys. Lett. 1998, 286,
253.
29. Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 106, 5151.
30. Ridley, J. E.; Zerner, M. C. J. Mol. Spectrosc. 1974, 50, 457.
31. Foresman, J. B.; Head-Gordon, M.; Pople, J. A.; Frisch, M. J. J. Phys. Chem.
1992, 96, 135.
32. Bauernschmitt, R.; Ahlrichs, R. Chem. Phys. Lett. 1996, 256, 454.
33. Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem. Phys.
1998, 108, 4439.
34. Yamaguchi, K.; Jensen, F.; Dorigo, A.; Houk, K. N. Chem. Phys. Lett. 1988,
149, 537.
35. Yamanaka, S.; Kawakami, T.; Nagao, H.; Yamaguchi, K. Chem. Phys. Lett.
1994, 231, 25.
36. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Montgomery, J., J. A.; Vreven, T.; Kudin, K. N.; Burant, J.
C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi,
M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara,
98
M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda,
Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.;
Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R.
E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala,
P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.;
Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.;
Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.;
Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.;
Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M.
W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A.
Gaussian 03, Gaussian, Inc.: Wallingford, CT, 2004.
37. Werner, H.-J.; Knowles, P. J. MOLPRO 2002, University of Birmingham:
Birmingham, 2002.
38. Goldstein, S.; Czapski, G.; Lind, J.; Merényi, G. Chem. Res. Toxicol. 2001, 14,
1273.
39. Lymar, S. V.; Hurst, J. K. Inorg. Chem. 1998, 37, 294.
40. Radi, R.; Peluffo, G.; Alvarez, M. N.; Naviliat, M.; Cayota, A. Free Radical
Biol. Med. 2001, 30, 463.
41. Bartberger, M. D.; Olson, L. P.; Houk, K. N. Chem. Res. Toxicol. 1998, 11,
710.
42. González Lebrero, M. C.; Perissinotti, L. L.; Estrin, D. A. J. Phys. Chem. A
2005, 109, 9598.
43. Armstrong, D. A.; Waltz, W. L.; Rauk, A. Can. J. Chem. 2006, 84, 1614.
44. Krauss, M. Chem. Phys. Lett. 1994, 222, 513.
99
45. Lo, W.-J.; Lee, Y.-P.; Tsai, J.-H. M.; Beckman, J. S. Chem. Phys. Lett. 1995,
242, 147.
46. Nakatsuji, H. Chem. Phys. Lett. 1978, 59, 362.
47. Nakatsuji, H. Chem. Phys. Lett. 1979, 67, 329.
48. Nakatsuji, H. Chem. Phys. Lett. 1979, 67, 334.
49. Czapski, G.; Ozeri, Y.; Goldstein, S. Radiat. Phys. Chem. 2005, 72, 229.
100
Chapter Five
Reaction of the Radical Pair NO2• and CO3•– with 2-[6-(4′-
Amino)phenoxy-3H-xanthen-3-on-9-yl]benzoic Acid (APF)
5.1 Introduction
Reactive nitrogen species (RNS) and reactive oxygen species (ROS) play numerous
and diverse roles in physiological pathways. They are capable of mediating nitrosative
and oxidative stress and exhibit both cytotoxic and cytoprotective effects [1]. The
RNS peroxynitrite (ONOO–) is a free radical precursor that is formed in cells by
diffusion-controlled reaction of nitric oxide (NO•) and superoxide (O2•–). The gas-
phase reaction enthalpy for the reaction of NO• with O2•– to form ONOO– at 298 K
was earlier calculated to be −138.3 kJ mol–1 at the CCSD(T)/cc-pVQZ//CCSD/cc-
pVTZ level [2]. It has been established recently that ONOO– reacts rapidly with
ubiquitous CO2 (k = 3 × 104 M–1 s–1) [3] to form nitrosoperoxycarbonate anions
(ONOOCO2–) which then undergo O–O bond homolysis to give NO2
• and CO3•–
radicals at a yield of about 30% [4-7]. These radicals can in turn react with nucleotide
bases in DNA and RNA, or amino acid residues in proteins, resulting in significant
structural changes and irreversible damage [8], although such reactions have not been
well characterized experimentally. It is widely believed that the action of these RNS
and ROS contributes to Alzheimer’s disease [9], Parksinson’s disease [10],
atherosclerosis [11] and cancer [12, 13].
101
The detection of RNS and ROS is a difficult task due to two main reasons: their short
lifespan, and their high reactivity with numerous other molecules. As a consequence,
detection, measurement and monitoring of RNS and ROS levels remain a constant
challenge to scientists. Fluorescent probes are highly sensitive and useful in
monitoring RNS and ROS. An ideal fluorescent probe should display high
fluorescence quantum yield for the species to be detected, and should be both
selective and accurate. At present, there are several fluorescent probes commercially
available for detecting and measuring ONOO–, for instance, 2′,7′-dicholorofluorescin
(DCFH), dihydrofluorescein (HFLUOR), dihydrorhodamine 123 (DHR), Rhodamine
B hydrazide, 2-[6-(4′-amino)phenoxy-3H-xanthen-3-on-9-yl]benzoic acid (APF) [14,
15] and the more recently developed HKGreen-1 [16]. In particular, Setsukinai et al
have shown that APF exhibits excellent selectivity for ONOO– over other radicals and
radical precursors such as NO•, O2•–, hydrogen peroxide (H2O2) and alkyl peroxide
radicals (ROO•), when compared with DCFH [17]. The APF molecule undergoes O-
dearylation in the presence of RNS generated by ONOO– derivatives to yield
fluorescein, with excitation and emission wavelengths of 490 and 520 nm,
respectively (Scheme 5.1). Although these probes are known to be oxidized to their
fluorescent forms in the presence of ONOO–, information on the exact identities of the
species that react with these probes is lacking.
102
Scheme 5.1. Action of RNS on APF, adapted from Ref.17
Experiments carried out by various research groups have shed some light on the
mechanism of fluorescent probe oxidation by ONOO– and its derivatives. The
exclusion of a direct reaction of ONOO– with a fluorescent probe was established by
Jourd’heuil et al [18]. The authors demonstrated that the oxidation of DHR by
ONOO– occurs via free radical intermediates NO2• and OH• formed from the
spontaneous decomposition of ONOOH. Kinetic studies by Koppenol et al reveal that
reaction of DCFH with ONOO– is essentially zero order with respect to the indicator,
eliminating the possibility of a direct reaction between ONOO– and the probe
molecule [19]. They proposed that the reactive species that oxidized DCFH is the
radical pair NO2• and CO3
•–.
Despite the fact that APF is used by many research groups to detect and measure
ONOO– in various experiments, both in vivo and in vitro, the reasons for which APF
exhibits selectivity for ONOO– over other RNS and ROS are still unclear. In this
chapter, a computational study of the reaction of APF with NO2• and CO3
•– is
presented, to shed some light into its reaction mechanism, and the factors that
influence the selectivity of the probe. An experimental examination of the influence
103
of HCO3– concentration on the ONOO– mediated two-electron oxidation of APF was
also carried out to better understand the chemistry pertinent to this intriguing reaction.
5.2 Methods
5.2.1 Computational Methods
A hybrid DFT method, the modified Perdew-Wang 1-parameter method for kinetics
(MPW1K), was employed in this study. Hybrid DFT methods have been shown to be
successful in obtaining accurate molecular geometries, vibrational frequencies and
bond energies [20]. The MPW1K functional was parameterized to yield improved
performance for kinetics calculations and bond energies as compared to other hybrid
DFT functionals such as B3LYP, BH&HLYP and mPWPW91 [21]. In particular, the
MPW1K functional yields improved accuracy in transition state geometries and
barrier heights compared to the abovementioned functionals [22], making the
MPW1K method an adequate choice for investigating the system in question.
Full geometry optimizations of all local minima and saddle points were carried out at
the MPW1K/6-31G(d) level of theory. Frequency calculations for all species were
carried out to verify if they were indeed local minima with all real frequencies, or
first-order saddle points (i.e. transition states) with one imaginary frequency. Closed-
and open-shell species were treated using the RHF and UHF formalisms, respectively.
Additional energy calculations were carried out at the MPW1K/6-311+G(d,p) level,
based on the MPW1K/6-31G(d) optimized geometry, to obtain a more accurate
104
estimate of the energetics of the reaction studied. The energies of open-shell species
were also spin corrected using the method developed by Yamaguchi et al [23, 24].
The polarizable continuum model of Tomasi et al [25-27] (PCM) was initially used to
investigate the effect of a solvent reaction field on the energetics of the reaction
studied. However, convergence problems were encountered in the optimization of
structures. Hence, geometry optimizations and frequency calculations were carried
out for selected local minima and transition states using the more robust Onsager
reaction field (SCRF=Dipole) method [28-30] at the MPW1K/6-31G(d) level. A
dielectric constant (ε) of 40.0 is used to model a polar medium. For a molecular ion
with a total charge Q, a Born charge term [31], –½(1 – 1/ε)Q2/a0 (where a0 is the
cavity radius), was added to the expression of solvation energy to account for the ion-
dipole interaction. It is important to note that the SCRF is not a suitable method to
model aqueous solvation where specific solvent-solute interaction is expected to be
important. Therefore the scope of this present solvation study is limited to an aprotic
polar solvent such as acetonitrile. Higher-level relative energies were obtained using
this solvation model at the MPW1K/6-311+G(d,p) level, together with spin correction
for open-shell species. All calculations were performed using the GAUSSIAN 03 [32]
program suite.
5.2.2 Experimental methods
APF {2-[6-(4’-amino)phenoxy-3H-xanthen-3-on-9-yl]benzoic acid} was purchased
from Alexis Corporation (Lausen, Switzerland). All other chemicals were obtained
from Sigma Ltd (St Louis, MO, USA), and freshly prepared on the day of the
105
experiment. Distilled water passed through a Millipore water purification system was
used for all purposes.
Sodium phosphate buffer (0.1 M) and sodium hydrogen carbonate solution (0.25 M)
were added along with 2 µl of stock APF solution into a 1 ml sample tube and
protected from light. This mixture is allowed to stand for about fifteen minutes to
allow equilibriation of hydrogen carbonate, carbon dioxide and carbonate in solution,
before other reagents were added. Hydrogen peroxide-free peroxynitrite was prepared
using the method described by Beckman et al [33], and added to the mixture in the
sample tube. Removal of residual H2O2 was confirmed using a commercial kit
(Amplex Red; Molecular Probes) [34]. Concentrations of ONOO– were redetermined
before each experiment using UV-visible spectroscopy, with ONOO– having a molar
extinction coefficient of 1670 cm–1, at 302 nm. ONOO– concentrations of 250 – 300
mM were usually obtained. This concentrated peroxynitrite (about 2 – 5 µl) was then
added to the reaction mixture in the sample tube. The sample tubes were then
vortexed and left to stand for five minutes before the fluorescence intensity was
measured. The pH of the well buffered reaction mixture upon adding of a small
amount of concentrated peroxynitrite is not expected to significantly change, as
already reported in previous work [35]. The final concentrations of the components in
the reaction mixture after mixing were: 10 µM APF, 1.0 µM ONOO–, 0.1 M
phosphate buffer, and varying concentrations of HCO3–, from 0 to 25 mM. The
fluorescence intensity that resulted from peroxynitrite-mediated oxidation of APF was
monitored with λexcitation = 490 nm and λemission = 520 nm at room temperature, using a
Gemini EM fluorescence microplate reader from Molecular Devices.
106
5.3 Results and Discussion
5.3.1 Influence of HCO3− concentration on ONOO
− mediated oxidation of APF
In an attempt to obtain some insight as to whether HCO3− substantially influence the
ONOO− mediated oxidation of APF, we studied the percentage yield of fluorescein as
a function of [HCO3–]. It was found that with increasing concentrations of HCO3
− in
solution, oxidation of APF was inhibited, though not completely (Figure 5.1). This
finding is consistent with the results obtained by Koppenol et al for the fluorogenic
indicators DCFH and DHR [19]. Increasing [HCO3–] shifts the equilibrium of the
reaction HCO3– = CO2 + OH
– further to the right, and the increased CO2 concentration
leads to the formation of larger amounts of ONOOCO2– from the reaction ONOO– +
CO2 = ONOOCO2–. The observation that increasing [HCO3
–] progressively lowers the
yield of fluorescent products instead of increasing it, appears to exclude the pathway
where APF reacts directly with ONOOCO2–. A similar conclusion can be reached, i.e.
the direct reaction between APF and ONOOCO2– is not likely.
107
HCO3– concentration / µM
15
20
25
30
35
40
0 2 4 6 8 10 12 14 16
%age yield of fluorescein, relative to ONOO–
HCO3– concentration / µM
15
20
25
30
35
40
0 2 4 6 8 10 12 14 16
%age yield of fluorescein, relative to ONOO–
Figure 5.1. Plot of percentage yield of fluorescein, relative to ONOO–, versus the concentration of HCO3
– in the reaction mixture. Other experimental conditions: 10 µM APF, 1.0 µM ONOO–, 0.1 M phosphate buffer. Total volume of the mixture is 1 ml.
The O—O bond dissociation energy (BDE) of ONOOCO2– was previously estimated
at the CBS-QB3 level to be 31.4 kJ mol–1 [36]. This magnitude of O—O BDE in
ONOOCO2– is unexpectedly low compared to O—O BDEs in other compounds such
as hydrogen peroxide HO–OH (213.4 kJ mol–1), dimethyl peroxide CH3O–OCH3
(158.6 kJ mol–1) and pernitric acid HO–ONO2 (163.2 kJ mol–1) [37]. The relative ease
of decomposition of ONOOCO2– supports the explanation proposed by Koppenol et al
for the DCFH and DHR case, that these fluorogenic indicators react with that the
decomposition products of ONOOCO2–, namely the radical pair NO2
• and CO3•– [19].
This explanation can be extended for the fluorogenic indicator APF.
5.3.2 Gas-Phase Calculations
The final product, fluorescein, can exist in a number of structural (quinoid,
zwitterionic and lactoid) and prototropic (neutral, monoanionic and dianionic) forms.
108
It exists purely as the dianionic form at pH 9.1 and the monoanionic form at pH 5.4.
At pH 7.4, fluorescein exists as a mixture of both forms [38]. Quantum mechanical
studies on the charge transfer transition for both forms show that the transition for the
dianion has markedly higher oscillator strength than that for the monoanion [39].
Thus, the monoanionic carboxylate form of APF, which yields dianionic fluorescein
after oxidation, was employed in this study.
Based on the two-electron oxidation products of APF reaction with RNS, namely
fluorescein and p-benzoquinone imine, and the reactive species NO2• and CO3
•–
involved as reactants, the reaction of APF can be interpreted as sequences of reactions
illustrated in Schemes 5.2 and 5.3. The reaction can occur via two plausible pathways.
The first, pathway I (Scheme 5.2), involves two rapid one-electron oxidations. The
first one-electron oxidation step involves a proton coupled electron transfer (PCET)
from the amino group in APF (1A) by NO2•. The second step involves O– addition by
CO3•– to form the resonance stabilized anilinyl fluorescein radical 1E, followed by
decomposition of the adduct 2B to yield fluorescein, 2E, p-benzoquinone imine, 2F,
and CO2. The second pathway, pathway II (Scheme 5.3), involves a direct addition of
CO3•– to 1A to give a radical adduct 3B. Decomposition of the adduct yields 2E, p-
aminophenoxyl radical, 3E, and CO2.
109
OR
NH
O2CO
OR
NH
O2COO
R
NH
OO2C
TS2A 2B TS2C
OR
NH
1E
OR
O
HN
CO2 NH
O
RO
2E
2F
+ + CO2
2D
OR
NH2
OR
HNH ONO
OR
HN H ONO
OR
HN HONO
OR
NH
OR
NH
1A 1B TS1C 1D 1E
R= OO
COO
CO3
NO2+ HONO
Scheme 5.2. Pathway I of APF oxidation.
OR
NH2
O2CO
OR
NH2
O2COO
R
NH2
O
TS3A 3B TS3C
OR
O
H2N
CO2
3D
OR
NH2
1A
NH2
O
RO
2E
3E
+ + CO2
R= OO
COO
O2C
CO3
Scheme 5.3. Pathway II of APF oxidation.
The calculated structural parameters of neutral fluorescein (Figure 5.2), are in good
accord with known experimental values [40], with maximum relative errors for bond
lengths at 5%. Calculated structural parameters of dianionic fluorescein 2E, are
essentially identical to those of the neutral form. A comparison of 11 known
vibrational modes of fluorescein [38] with the calculated MPW1K/6-31G(d)
frequencies yielded errors between 4−18%. However, upon applying a frequency
110
scaling factor of 0.905, the relative errors were reduced to 1.1−5.7%. The relatively
good agreement of calculated geometries and frequencies with experimental values
provided the confidence that the choice of the MPW1K/6-31G(d) method is adequate
for the study of the reaction in question. With the lack of reliable zero-point energy
data for fluorescein, the scaling factor for fundamentals was applied to calculated the
zero-point vibrational energies.
1.219
[1.219]
(1.222)
1.469
[1.469]
(1.398)
1.451
[1.451]
(1.415)
1.344
[1.345]
(1.402)
1.355
[1.355]
(1.364)1.446
[1.446]
(1.398)
1.339
[1.339]
(1.412)
1.442
[1.442]
(1.410)
1.360
[1.360]
(1.414)
1.437
[1.437]
(1.398)
1.342
[1.342]
(1.356)
1.399
[1.399]
(1.407)
1.381
[1.381]
(1.407)
1.384
[1.384]
(1.392)
1.342
[1.343]
(1.365)1.397
[1.397]
(1.409)
1.397
[1.397]
(1.411)
1.375
[1.375]
(1.410)
1.486
[1.486]
(1.495)1.391
[1.391]
(1.410)
1.384
[1.384]
(1.406)
1.385
[1.385]
(1.400)
1.380
[1.380]
(1.400)
1.392
[1.392]
(1.415)
1.401
[1.401]
(1.401)
1.485
[1.486]
(1.511)
1.329
[1.330]
(1.339)
1.201
[1.202]
(1.235)
1.219
[1.219]
(1.222)
1.469
[1.469]
(1.398)
1.451
[1.451]
(1.415)
1.344
[1.345]
(1.402)
1.355
[1.355]
(1.364)1.446
[1.446]
(1.398)
1.339
[1.339]
(1.412)
1.442
[1.442]
(1.410)
1.360
[1.360]
(1.414)
1.437
[1.437]
(1.398)
1.342
[1.342]
(1.356)
1.399
[1.399]
(1.407)
1.381
[1.381]
(1.407)
1.384
[1.384]
(1.392)
1.342
[1.343]
(1.365)1.397
[1.397]
(1.409)
1.397
[1.397]
(1.411)
1.375
[1.375]
(1.410)
1.486
[1.486]
(1.495)1.391
[1.391]
(1.410)
1.384
[1.384]
(1.406)
1.385
[1.385]
(1.400)
1.380
[1.380]
(1.400)
1.392
[1.392]
(1.415)
1.401
[1.401]
(1.401)
1.485
[1.486]
(1.511)
1.329
[1.330]
(1.339)
1.201
[1.202]
(1.235)
Figure 5.2. Calculated structural parameters of dianionic fluorescein, 2E, at the MPW1K/6-31G(d) level in the gas phase. Calculated structural parameters for neutral fluorescein at the MPW1K/6-31G(d) level are given in square brackets. X-ray structural parameters of neutral fluorescein (from Ref. 38) are given in parentheses. Non essential hydrogens are omitted for clarity. Distances are in Ångstroms. Oxygen atoms are in dark color.
Gas-phase transition energy (λexcitation) for dianionic fluorescein was calculated using
ZINDO/S [41] and TD-DFT [42, 43] methods. The predicted λexcitation using the
ZINDO/S method is 475 nm, in close agreement with the experimental value of 490
nm [44]. TD-B3LYP/6-311+G(d,p)//B3LYP/6-31G(d) and TD-MPW1K/6-
111
311+G(d,p)//MPW1K/6-31G(d) predict λexcitation values of 430 nm and 392 nm,
respectively. It is important to note that time-dependent density-functional theory
yields unsatisfactory results for excitation energies of long-range charge-transfer
states [45].
Schematic energy profiles of the APF oxidation reactions in the gas phase, calculated
at MPW1K/6-311+G(d,p)//MPW1K/6-31G(d) + ZPE level, are presented in Figures
5.3 to 5.5. Gas-phase MPW1K/6-31G(d) optimized structures of selected local
minima (1A, 1B, 1D, 1E, 2B, 2D, 3B, 3D and 3E) and all transition states (TS1C,
TS2A, TS2C, TS3A and TS3C) are presented in Figures 5.6 to 5.8. The atom
numbering scheme used for 1A, 1E, TS2C and TS3C is given in Figure 5.9.
-10
0
10
20
30
40
50
60
70
80
90
1A + NO2•0.0
1B
-65.4
TS1C
-18.9
1D
-35.7
1E + HONO18.2
Energy (kJ mol−1 )
Figure 5.3. Schematic gas-phase reaction energy profile for step 1 of pathway I, calculated at the MPW1K/6-311+G(d,p)//MPW1K/6-31G(d) +ZPE level with spin correction.
In pathway I, NO2• first interacts with 1A to form a hydrogen-bonded reactant
complex 1B, with a relatively large binding energy of 65.4 kJ mol–1. The hydrogen
112
transfer reaction involves the transfer of the amino hydrogen of 1B to NO2•. It occurs
via TS1C, with an activation barrier of 46.5 kJ mol–1. The N…H and O…H bond
distances in TS1C are 1.143 and 1.367 Å, respectively. The N…H…O moiety is close
to linearity (Figure 5.8). Because of the larger interaction of the pre-transition state
complex 1B, the hydrogen transfer reaction has a negative barrier with respect to 1A
+ NO2•. The hydrogen bonded product complex 1D is slightly less stable than the
reactant complex 1B by 29.7 kJ mol–1.
It is important to note that the carbonate radical anion (CO3•–) is a strong one-electron
oxidant that oxidizes appropriate electron donors via electron transfer mechanisms
[46-48]. The redox potential of CO3•– (equation 5-1) is well established [49, 50] and is
strongly positive, 1.6 (pH 12.5) to 1.78 V (pH 7.0) vs. NHE. In fact, this value is
greater than the redox potential of NO2• (equation 5-2) [51].
CO3•– + e– + H+ → HCO3
– E° = 1.78 V vs. NHE (5-1)
NO2• + e– → NO2
– E° ≈ 1.04 V vs. NHE (5-2)
Thus, the first one-electron oxidation step of pathway I may involve CO3•– (equation
7-3).
1A + CO3•– → 1E + HCO3
– (5-3)
The proton coupled electron transfer (PCET) from the amino group in APF (1A) by
CO3•– will also lead to the formation of 1E together with HCO3
–, a fairly stable
bicarbonate anion. The calculated reaction energy of this one-electron oxidation
reaction is exothermic, by 45.0 kJ mol–1. However, this electron transfer reaction is
expected to be unfavorable kinetically as the transition state involves two anions
113
approaching each other. Indeed, this proton coupled electron transfer reaction is
calculated to have a large barrier of 106.9 kJ mol–1. The dianionic transition state is
significantly higher in energy (by 125.8 kJ mol–1) compared to the transition state
involving NO2• (TS1C). This result suggests that the alternate one-electron oxidation
via carbon radical anion is less viable.
-100
-50
0
50
100
150
200
1E + CO3•−
0.0
TS2A
162.2
2B
28.7
TS2C
36.4
2D
-40.2
2E + 2F + CO2
61.8
Energy (kJ mol−1 )
Figure 5.4. Schematic gas-phase reaction energy profile for step 2 of pathway I, calculated at the MPW1K/6-311+G(d,p)//MPW1K/6-31G(d) +ZPE level with spin correction.
The CO3•− radical can add to 1E in three plausible positions: C(23), C(25) and C(27)
(See Figure 5.9). The only position involving CO3•− radical addition to yield the final
products is at C(23) and, hence, is the only one considered here. Addition of the CO3•−
radical to 1E to form adduct 2B is endothermic by 28.7 kJ mol–1, and proceeds via a
transition state TS2A with strong biradical character. The forming C…O bond length
in TS2A is 2.035 Å. The expectation value of the spin-squared operator ⟩⟨2S of
TS2A is 0.742, indicating that there is a significant degree of spin contamination from
the higher energy spin states. The activation barrier for CO3•− addition via TS2A
before spin correction is 193.0 kJ mol–1. After the addition of the spin correction term,
this barrier is reduced significantly to 162.2 kJ mol–1.
114
Decomposition of 2B to form 2E, 2F and CO2 is endothermic by 33.1 kJ mol–1 and
takes place via TS2C. Two C—O bonds break simultaneously, between C(23) and
O(22) and between C(31) and O(30) in TS2C (See Figure 5.8). C(23)—O(22) and
C(31)—O(30) distances are 1.853 Å and 2.521 Å, respectively in TS2C. Remarkably
this fragmentation process is inhibited by a small barrier of just 7.7 kJ mol–1. As
expected, the final products form a stable complex 2D (2E…2F
…CO2) with a fairly
large binding energy of 102.0 kJ mol–1.
-100
-50
0
50
100
150
200
250
1A + CO3•−
0.0
TS3A
233.5
3B
132.4
TS3C
150.1
3D
-44.6
2E+3E+CO2
80.8
Energy (kJ mol−1 )
Figure 5.5. Schematic gas-phase reaction energy profile for pathway II, calculated at the MPW1K/6-311+G(d,p)//MPW1K/6-31G(d) +ZPE level with spin correction.
The radical addition process via pathway II is similar to the second step of pathway I.
The CO3•− radical is expected to undergo radical addition with the initial reactant 1A
instead of the intermediate formed after hydrogen transfer. As with pathway I, the site
considered for CO3•− addition is at C(23). The direct addition of CO3
•− radical anion to
1A to form 3B is highly endothermic, by 132.4 kJ mol–1. The activation barrier for
this process is notably high at 233.5 kJ mol–1, via TS3A. The transition state of this
115
reaction, TS3C, has similar structural characteristics as TS2C of pathway I. The
C(23)—O(22) and C(31)—O(30) distances in TS3C are comparatively shorter, at
1.685 and 2.239 Å respectively. The products 2E, 3E and CO2 form a complex with
substantial binding energy of 125.4 kJ mol–1. Overall, addition of CO3•− to 1A to form
3B is the rate determining step of pathway II.
1A 1B
1D 1E
1.769
1.710
N
NN
N
N
N
Figure 5.6. MPW1K/6-31G(d) optimized geometries of 1A, 1B, 1D and 1E. Distances shown are in Ångstroms. Non essential hydrogens are omitted for clarity. Oxygen atoms are in dark color.
116
2B 2D
3B 3D
1.3621.788
2.673
1.369
1.558
2.559
N
N
N
N
Figure 5.7. MPW1K/6-31G(d) optimized geometries of 2B, 2D, 3B and 3D. Distances shown are in Ångstroms. Non essential hydrogens are omitted for clarity. Oxygen atoms are in dark color.
117
TS1C
TS2A TS2C
TS3CTS3A
1.143
1.367
2.035
1.853
2.521
1.875
1.685
2.239
N
N
N N
NN
Figure 5.8. MPW1K/6-31G(d) optimized geometries of transition structures TS1C, TS2A, TS2C, TS3A and TS3C. Distances shown are in Ångstroms. Non essential hydrogens are omitted for clarity. Oxygen atoms are in dark color.
118
Figure 5.9. Atom numbering scheme for 1A, 1E, TS2C and TS3C used in this study.
Based on the calculated gas-phase potential energy surface, oxidation of APF to form
fluorescein can take place via two pathways. Pathway I involves initial proton
coupled electron transfer (PCET) from APF to NO2•, followed by O•– addition of
CO3•− to the radical 1E, which, with a barrier of 162.2 kJ mol–1, is the rate limiting
step. Pathway II is independent of NO2•, but is not likely to proceed due to a
prohibitively high barrier of 233.5 kJ mol–1 for the rate determining step.
5.3.3 Solvent Effects
To investigate the influence of a solvent reaction field on the reaction mechanism,
119
species 1A, 1B, 1E, 2B, 3B, TS1C, TS2A, TS2C, TS3A and TS3C were re-
optimized at the MPW1K/6-31G(d) level incorporated with the Onsager reaction field
model with a dielectric medium of ε = 40.0. The MPW1K/6-311+G(d,p)//MPW1K/6-
31G(d) activation barriers corresponding to 1B → TS1C, 1E + CO3•− → TS2A, 2B
→ TS2C, 1A + CO3•− → TS3A, and 3B → TS3C in the solvent medium are
presented in Table 5.1, with the gas-phase MPW1K/6-311+G(d,p)//MPW1K/6-31G(d)
barriers shown alongside for comparison.
For pathway I, activation barriers 1B → TS1C and 1E + CO3•− → TS2A decreased by
19.2 and 101.3 kJ mol–1, respectively, in a polar dielectric medium. The vast decrease
in barrier height for 1E + CO3•− → TS2A is due mainly to the additional contribution
of an ion-dipole (Born charge) term in the dianionic transition state TS2A. The barrier
for 2B → TS2C in the polar dielectric medium, on the other hand, increased by 36.4
kJ mol–1. The rate limiting step is still the addition of CO3•− radical to 1E. However,
APF oxidation via pathway I is generally more favorable in solvent than in the gas
phase, with barriers ranging from 27.2 to 60.9 kJ mol–1.
Reaction Gas Phase (ε =1) Polar medium (ε = 40)a
Pathway I
1B → TS1C 46.4 27.2
1E + CO3•− → TS2A 162.2 60.9
2B → TS2C 7.7 44.1
Pathway II
1A + CO3•− → TS3A 233.5 218.2
3B → TS3C 17.7 19.5
Table 5.1. Calculated activation barriers (kJ mol–1) of the two reaction pathways in the gas phase and in a polar medium at the MPW1K/6-311+G(d,p)//MPW1K/6-31G(d) level. a SCRF=dipole calculations.
120
For pathway II, the activation barrier 1A + CO3•− → TS3A is lowered by 15.3 kJ mol–
1, and the barrier 3B → TS3C is slightly increased by 1.8 kJ mol–1. Comparing with
pathway I, pathway II is less likely to proceed because of the prohibitively high
barrier of CO3•− radical addition to 1A.
5.4 Concluding Remarks
The reaction of APF with the radicals generated by the reaction of ONOO– and CO2,
namely NO2• and CO3
•−, was explored. Experimentally, HCO3– was found to inhibit
the two-electron oxidation of APF, suggesting that the direct reaction of APF with
ONOOCO2– is unlikely. Two possible pathways of APF oxidation were explored
using DFT calculations. Pathway I involves initial proton coupled electron transfer
from APF to NO2 to obtain a resonance stabilized anilinyl fluorescein radical. CO3•−
addition to this radical follows, and subsequently decomposition of the adduct yields
fluorescein, p-benzoquinone imine and CO2. Pathway II involves the direct addition
of CO3•− to APF, and the subsequent decomposition of the APF-CO3
•− adduct to form
fluorescein, p-aminophenoxyl radical and CO2. Addition of CO3•− is the rate
determining step in both pathways, and the activation barrier for this step in pathway
II is prohibitively high even in a polar aprotic solvent. Hence, the two-electron
oxidation of APF by is likely to involve two one-electron oxidations by both NO2• and
CO3•− radicals and proceeds via pathway I.
121
References
1. Wink, D. A.; Mitchell, J. B. Free Radical Biol. Med. 1998, 25, 434.
2. Mak, A. M.; Wong, M. W. Chem. Phys. Lett. 2005, 403, 192.
3. Lymar, S. V.; Hurst, J. K. J. Am. Chem. Soc. 1995, 117, 8867.
4. Bonini, M. G.; Radi, R.; Ferrer-Sueta, G.; Ferreira, A. M. D. C.; Augusto, O. J.
Biol. Chem. 1999, 274, 10802.
5. Goldstein, S.; Czapski, G.; Lind, J.; Merényi, G. Chem. Res. Toxicol. 2001, 14,
1273.
6. Radi, R.; Peluffo, G.; Alvarez, M. N.; Naviliat, M.; Cayota, A. Free Radical
Biol. Med. 2001, 30, 463.
7. Squadrito, G. L.; Pryor, W. A. Free Radical Biol. Med. 1998, 25, 392.
8. Augusto, O.; Bonini, M. G.; Amanso, A. M.; Linares, E.; Santos, C. X. C.; de
Menezes, S. L. Free Radical Biol. Med. 2002, 32, 841.
9. Butterfield, D. A. Free Radical Res. 2002, 36, 1307.
10. Riobó, N. A.; Schöpfer, F. J.; Boveris, A. D.; Cadenas, E.; Poderoso, J. J. Free
Radical Biol. Med. 2002, 32, 115.
11. Halliwell, B. Cardiovasc. Res. 2000, 47, 410.
12. Hagen, T. M.; Huang, S.; Curnutte, J.; Fowler, P.; Martinez, V.; Wehr, C. M.;
Ames, B. N.; Chisari, F. V. Proc. Natl. Acad. Sci. U. S. A. 1994, 91, 12808.
13. Wink, D. A.; Mitchell, J. B. Free Radical Biol. Med. 2003, 34, 951.
14. Gomes, A.; Fernandes, E.; Lima, J. L. F. C. J. Fluoresc. 2006, 16, 119.
15. Halliwell, B.; Whiteman, M. Br. J. Pharmacol. 2004, 142, 231.
16. Yang, D.; Wang, H.-L.; Sun, Z.-N.; Chung, N.-W.; Shen, J.-G. J. Am. Chem.
Soc. 2006, 128, 6004.
122
17. Setsukinai, K.-i.; Urano, Y.; Kakinuma, K.; Majima, H. J.; Nagano, T. J. Biol.
Chem. 2003, 278, 3170.
18. Jourd'heuil, D.; Jourd'heuil, F. L.; Kutchukian, P. S.; Musah, R. A.; Wink, D.
A.; Grisham, M. B. J. Biol. Chem. 2001, 276, 28799.
19. Glebska, J.; Koppenol, W. H. Free Radical Biol. Med. 2003, 35, 676.
20. Raghavachari, K. Theor. Chem. Acc. 2000, 103, 361.
21. Lynch, B. J.; Fast, P. L.; Harris, M.; Truhlar, D. G. J. Phys. Chem. A 2000,
104, 4811.
22. Lynch, B. J.; Truhlar, D. G. J. Phys. Chem. A 2001, 105, 2936.
23. Yamaguchi, K.; Jensen, F.; Dorigo, A.; Houk, K. N. Chem. Phys. Lett. 1988,
149, 537.
24. Yamanaka, S.; Kawakami, T.; Nagao, H.; Yamaguchi, K. Chem. Phys. Lett.
1994, 231, 25.
25. Cancès, E.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107, 3032.
26. Cossi, M.; Barone, V.; Mennucci, B.; Tomasi, J. Chem. Phys. Lett. 1998, 286,
253.
27. Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 106, 5151.
28. Wong, M. W.; Wiberg, K. B.; Frisch, M. J. Chem. Phys. 1991, 95, 8991.
29. Wong, M. W.; Wiberg, K. B.; Frisch, M. J. J. Am. Chem. Soc. 1991, 113, 4776.
30. Wong, M. W.; Wiberg, K. B.; Frisch, M. J. J. Am. Chem. Soc. 1992, 114, 1645.
31. Born, M. Z. Phys. 1920, 1, 45.
32. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Montgomery, J., J. A.; Vreven, T.; Kudin, K. N.; Burant, J.
C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi,
M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara,
123
M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda,
Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.;
Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R.
E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala,
P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.;
Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.;
Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.;
Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.;
Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M.
W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A.
Gaussian 03, Gaussian, Inc.: Wallingford, CT, 2004.
33. Beckman, J. S.; Chen, J.; Ischiropoulos, H.; Crow, J. P., Oxidative Chemistry
of Peroxynitrite. In Methods Enzymol., Packer, L., Ed. Academic Press: San
Diego, 1994; Vol. 233, pp 229.
34. Whiteman, M.; Armstrong, J. S.; Cheung, N. S.; Siau, J.-L.; Rose, P.; Schantz,
J.-T.; Jones, D. P.; Halliwell, B. FASEB J. 2004, 18, 1395.
35. Whiteman, M.; Szabó, C.; Halliwell, B. Br. J. Pharmacol. 1999, 126, 1646.
36. Olson, L. P.; Bartberger, M. D.; Houk, K. N. J. Am. Chem. Soc. 2003, 125,
3999.
37. Luo, Y.-R., Handbook of Bond Dissociation Energies in Organic Compounds.
CRC Press: Boca Raton, FL, 2003.
38. Wang, L.; Roitberg, A.; Meuse, C.; Gaigalas, A. K. Spectrochim. Acta, Part A
2001, 57, 1781.
124
39. Tamulis, A.; Tamuliene, J.; Balevicius, M. L.; Rinkevicius, Z.; Tamulis, V.
Struct. Chem. 2003, 14, 643.
40. Tremayne, M.; Kariyuki, B. M.; Harris, K. D. M. Angew. Chem. Int. Ed. 1997,
36, 770.
41. Ridley, J. E.; Zerner, M. C. J. Mol. Spectrosc. 1974, 50, 457.
42. Bauernschmitt, R.; Ahlrichs, R. Chem. Phys. Lett. 1996, 256, 454.
43. Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem. Phys.
1998, 108, 4439.
44. Parker, C. A.; Rees, W. T. Analyst (Cambridge, U. K.) 1960, 85, 587.
45. Dreuw, A.; Weisman, J.; Head-Gordon, M. J. Chem. Phys. 2003, 119, 2943.
46. Bonini, M. G.; Augusto, O. J. Biol. Chem. 2001, 276, 9749.
47. Ferrer-Sueta, G.; Vitturi, D.; Batinić-Haberle, I.; Fridovich, I.; Goldstein, S.;
Czapski, G.; Radi, R. J. Biol. Chem. 2003, 278, 27432.
48. Armstrong, D. A.; Waltz, W. L.; Rauk, A. Can. J. Chem. 2006, 84, 1614.
49. Czapski, G.; Lymar, S. V.; Schwarz, H. A. J. Phys. Chem. A 1999, 103, 3447.
50. Huie, R. E.; Clifton, C. L.; Neta, P. Radiat. Phys. Chem. 1991, 38, 477.
51. Stanbury, D. Adv. Inorg. Chem. 1989, 33, 69.
125
Chapter Six
Homolytic Dissociation of the Vulcanization Accelerator
Tetramethylthiuram Disulfide (TMTD) and Structures and
Stabilities of the Related Radicals Me2NCSn• (n = 1−4)
6.1 Introduction
The global annual production of rubber is approximately 17 million tons. In order to
turn natural or synthetic polyolefins or “rubbers” into useful elastomers, sulfur
vulcanization is usually performed to generate sulfidic and disulfidic cross-links
between neighboring macromolecules. Tetramethylthiuram disulfide (TMTD) is one
of the most important accelerators of this type of vulcanization (see Figure 6.1).
Figure 6.1. Chemical structure of tetramethylthiuram disulfide (TMTD)
TMTD is usually applied together with zinc oxide, and the function of the latter in this
context has been studied extensively [1-3]. The formation of zinc thiolate complexes
has been postulated, which are believed to add more sulfur atoms either from each
other by disproportionation or from added elemental sulfur. These polysulfido
126
complexes are assumed to attack the polyolefin of natural or synthetic rubbers,
eventually resulting in cross-linking between polymeric chain molecules [4, 5].
The thermal behavior of pure TMTD has repeatedly been studied by several groups.
Free radicals have been detected by electron spin resonance (ESR) spectroscopy when
TMTD was heated to 130–150°C [6, 7]. The intriguing question here is which of the
covalent bonds of TMTD was broken homolytically. A triplet signal at g = 2.03 with a
peak separation of ca. 15 G was observed after heating TMTD followed by cooling to
room temperature. This signal was assigned to the thiuram radical Me2NCS2•. A
singlet signal at g = 2.02 was observed when TMTD was heated and subsequently
quenched to –56°C. This signal was assigned to the persulfenyl radical Me2NCS3•. In
addition, a weak signal at g = 2.006 was recorded after heating TMTD for 2 hours to
145°C; this signal was assigned to an unknown carbon radical [6]. All these
assignments have not been substantiated by theoretical calculations as yet, and the
structures of various Me2NCSn• species except for the thiuram radical Me2NCS2
• are
as yet unknown. The structure of Me2NCS2• was investigated using MNDO RHF-CI
calculations by Plyusnin et al [8]. A planar C2NCS2 skeleton was predicted with two
identical C—S bonds 1.61 Å in length.
Nichols et al generated the radical iPr2NCS2• by thermolysis of the related
tetraisopropylthiuram disulfide [9]. Detection of this radical was achieved by ESR
spectroscopy. The precursor molecule was first dissolved in decalin and heated to
temperatures above 45°C. A single line at g = 2.015 was observed in the ESR
spectrum. From the temperature dependence of the ESR signal intensity, the
dissociation enthalpy of (iPr2NCS2)2 was calculated at 104 ± 2 kJ mol–1. The
127
dissociation enthalpy of the corresponding methyl derivative (TMTD) was roughly
estimated to lie in the range of 100–130 kJ mol–1 [9]. The iPr2NCS2• radical generated
using laser flash photolysis from the corresponding disulfide with acetonitrile as a
solvent showed an absorption band at 580 nm [8].
Depending on the nature and concentration of the Me2NCSn• radicals, the
vulcanization of rubber accelerated by TMTD can take place by either an anionic or a
radical mechanism or by both. Under equilibrium conditions, the concentration of
radicals depends on the various dissociation enthalpies of TMTD, which are not
known accurately. To address this question, the structures of the radicals Me2NCSn• (n
= 1–4) have been calculated and the various homolytic dissociation enthalpies of
TMTD have been studied using high-level ab initio calculations according to the
G3X(MP2) and G3X(MP2)-RAD theories. In addition, the influence of a metal cation
on the dissociation was investigated using lithium ions as a probe and as a model ion
for the impact of zinc cations during rubber vulcanization accelerated by the
TMTD/ZnO system. The zinc ions on the surface of solid ZnO particles have an
atomic charge much smaller than +2. Their interaction with TMTD and other
substrates can therefore be modeled by a singly charged ion such as Li+.
6.2 Computational Methods
The structures and energies of TMTD and its various derivatives were examined using
G3X(MP2) theory [10]. This composite method corresponds effectively to
QCISD(T)/G3XLarge//B3LYP/6-31G(2df,p) energy calculations together with zero-
128
point vibrational and isogyric corrections. The G3X(MP2) theory represents a
modification of G3(MP2) [11] theory with three important changes: (1) B3LYP/6-
31G(2df,p) instead of MP2(full)/6-31G(d) geometry, (2) B3LYP/6-31G(2df,p) instead
of HF/6-31G(d) zero-point vibrational energy, and (3) the addition of a g polarization
function to the G3Large basis set for second-row atoms at the Hartree-Fock level. All
three modifications are particularly important for the proper description of sulfur-
containing species. Harmonic fundamental vibrations were calculated at the
B3LYP/6-31G(2df,p) level to characterize stationary points as equilibrium structures
with all real frequencies and transition states with one and only one imaginary
frequency. Selected reactions were also examined using G3X(MP2)-RAD theory [12],
which was designed to provide better theoretical treatment for free radicals. The RHF
and UHF formalisms were employed for closed-shell and open-shell systems,
respectively, except for certain G3X(MP2)-RAD calculations.
For all investigated species, a charge density analysis was performed using the natural
bond orbital (NBO) [13] approach based on the B3LYP/6-31G(2df,p) wave function.
NBO atomic charges of small molecules have recently been demonstrated to agree
well with experimental values obtained from X-ray diffraction data [14]. Unless
otherwise noted, relative energies reported in this chapter correspond to G3X(MP2)
∆H°298 values. The same temperature of 298 K holds for the reported reaction
enthalpies and Gibbs free energies. All reported structural parameters correspond to
those obtained at the B3LYP/6-31G(2df,p) level. The electronic transition energies of
the Me2NCS2• radical and its isopropyl analogue were performed using the time-
dependent DFT (TD-DFT) method [15] at the B3LYP/6-31G(2df,p) level. The DFT
calculations of ESR g-tensors and g-values were based on the gauche-including
129
atomic orbitals (GIAO) approach [16-18] and were carried out using the Amsterdam
density functional package ADF [19-21]. All other calculations were performed using
the GAUSSIAN 03 [22] and MOLPRO 2002 [23] programs.
6.3 Results and Discussion
6.3.1 Structures and Stabilities of Tetramethylthiuram Sulfides and Their Related
Radicals
The calculated structure of the TMTD molecule is shown in Figure 6.2, together with
the structural data obtained from X-ray diffraction on single crystals at 138 K. Solid
TMTD crystallizes as two allotropes: a low-temperature phase of unknown structure
below 133 K and a monoclinic high-temperature phase in which the two independent
molecules of the asymmetric unit are of C2 symmetry [24]. The most stable calculated
structure 1 is also of C2 symmetry. Calculated and experimental bond lengths and
generally agree well to within 0.03 Å. There exists a rotational isomer of 1 (1a) that is
less stable by 13.9 kJ mol–1. The main difference between 1 and 1a are the N–C–S1–
S2 and the N–C–S3–S4 improper torsional angles. These angles are both –177.3° in 1,
compared to 9.5° and 155.0° in 1a. The C–S2–S3–C torsional angle is 88.7° in 1 and
78.3° in 1a. It is worth noting that the C—N bonds in 1 and 1a are relatively short
(1.332 – 1.347 Å) compared to a typical C—N bond length of 1.46 Å. This indicates
some extent of delocalization of the π electrons in the NCS frameworks.
130
1.347
(1.340)
1.840
(1.809)
2.023
(2.019)
1.650
(1.653)
1.347
(1.340)
1.840
(1.809)
2.023
(2.019)
1.650
(1.653) 1.3491.658
1.821
2.105
1.791
1.666
1.332
1.3491.658
1.821
2.105
1.791
1.666
1.332
1 (C2) 1a (C1)
S1
S2
S3
S4
Figure 6.2. B3LYP/6-31G(2df,p) optimized geometries of the two conformers of TMTD. Distances are in Ångstroms. X-ray structural parameters are given in parentheses.
Previous authors have discussed the possibilities that on heating TMTD dissociates
symmetrically (S—S bond cleavage) giving two Me2NCS2• radicals (2) or
asymmetrically (C—S bond cleavage) yielding the radicals Me2NCS• (3) and
Me2NCS3• (4) [6]. The calculated structures of these radicals are shown in Figure 6.3.
It is interesting to note that the structural units C2NCSn (n = 1–3) are either exactly
planar (2 and 3) or nearly planar (4). Expectedly, the most dramatic structural changes
on dissociation of TMTD take place at the CS2 unit, particularly with the bond lengths.
In TMTD there are two regular C=S double bonds (1.650 Å) and two C—S single
bonds (1.840 Å). Radical 3 has just one very short C=S bond of length 1.610 Å. Also,
the C—N bond length has decreased from 1.347 Å in 1 to 1.310 Å in 3. Evidently, the
π electron system of 3 extends from sulfur to nitrogen. In contrast, radical 2 has two
identical C—S bonds (1.710 Å) and a relatively short C—N bond (1.335 Å).
Structurally, the persulfenyl radical 4 resembles the parent compound 1.
131
1.456 1.335 1.700 1.467
1.461
1.3101.610 1.459
1.460
1.342
1.698
1.738
2.075
2 (C2v, 2B2) 3 (Cs,
2A′) 4 (C1, 2A)
S1 S1
S1
S2
S3
S2
1.456 1.335 1.700 1.467
1.461
1.3101.610 1.459
1.460
1.342
1.698
1.738
2.075
2 (C2v, 2B2) 3 (Cs,
2A′) 4 (C1, 2A)
S1 S1
S1
S2
S3
S2
Figure 6.3. B3LYP/6-31G(2df,p) optimized geometries of the radicals Me2NCSn• (n = 1–3). Distances
are in Ångstroms. Molecular symmetry and electronic states are given in parentheses.
It can be seen, from spin density plots in Figure 6.4, that all three radicals 2, 3 and 4
correspond to a σ-type radical. For radical 2 (2B2 ground electronic state), the
unpaired spin density is located almost exclusively in the two sulfur atoms.
Essentially, the SOMO is a linear combination of the two sulfur lone pair orbitals. The
resonance hybrids of 2 clearly show the delocalization of the π orbitals over the four-
center NCS2 framework as well as the delocalization of the spin density (see Figure
6.5).
132
Figure 6.4. UB3LYP/6-31G(2df,p) spin density plots of the Me2NCSn• radicals (n = 1–3)
Figure 6.5. Chemical drawing of 2, showing resonance structures.
The radical 2 has a large dipole moment of 4.92 D as expected from contributing
dipolar resonance forms. For comparison, the parent radical HCS2• has a smaller
dipole moment of 1.49 D at the same level of theory. It is noted that the π-type radical
(2A2) of 2 corresponds to an excited state with a relative enthalpy of 71.3 kJ mol–1.
The stability of this radical is readily reflected in the S—H bond dissociation enthalpy
133
(BDE) of the corresponding acid Me2NC(=S)SH. The calculated value of 301.7 kJ
mol–1 is the lowest among various sulfur-containing compounds as the following
examples demonstrate: 381.6 (H2S), 365.7 (CH3SH), 349.4 (PhSH), 330.5 (HOSH),
and 318.0 (HSSH) kJ mol–1 [25]. The stabilizing effect of the dimethylamino group in
radical 2 can be estimated by the following isodesmic reaction:
Me3N + HCS2• → Me2NCS2
• + CH4
∆H°298 = −159.0 kJ mol–1
(1)
The computed radical stabilization energy is substantially larger than those calculated
for carbon radicals XCH2• [26]. As in the case of 2, the spin density of the radical 4
resides mainly in the two terminal sulfur atoms. In the case of 3, however, the
unpaired electron spin is located in both the carbon and sulfur atoms (see Figure 6.4).
The ESR g-values of radicals 2−4 were investigated by the DFT-GIAO method at the
BLYP using Slater-type orbitals of the TZ2P basis set. The calculated isotropic g-
values are 2.017, 2.001 and 2.017 for 2, 3, and 4, respectively. Both radicals 2 and 4
are predicted to have similar g-values and would yield triplet signals on the ESR
spectrum, as the lone electron resides on the sulfur atoms which are β from the
nitrogen atom having nuclear spin I = 1 which causes peak splitting in the ratio 1:1:1.
The question of whether radical 2 or 4 is responsible for the observed ESR signal
arises. Homolytic cleavage of the S—S bond in TMTD yields two molecules of 2.
Similarly, the cleavage of the C—S bond in TMTD would yield 3 and 4. As predicted
from the calculated isotropic g-values, a mixture of 3 and 4 would show up as two
134
distinct triplets with g-values of 2.001 and 2.017, respectively. It is thus safe to assign
the observed ESR signal to the radical 2. The homolytic dissociation energies of the
S—S bond and of one of the C—S bonds in TMTD further provides evidence for this
claim. S—S and C—S bond dissociation energies differ considerably by 111.3 kJ
mol–1 shown in equations (2) and (3). The more favorable process yields radical 2 as
the product.
(Me2NCS2)2 (1) → 2Me2NCS2• (2)
∆H°298 = 155.1 kJ mol–1 ∆G°298 = 100.3 kJ mol–1
(2)
(Me2NCS2)2 (1) → Me2NCS• (3) + Me2NCS3
• (4)
∆H°298 = 266.4 kJ mol–1 ∆G°298 = 206.3 kJ mol–1
(3)
The calculated g-value of 3 appears to match well with the ESR signal previously
assigned to an “unknown carbon radical” [6]. The isotropic g-value of the radical
iPr2NCS2•, which has a definitive ESR spectrum, was computed to be 2.015, in perfect
agreement with the observed g-value. This computed g-value for iPr2NCS2• is very
close to that calculated for the methyl analogue (g = 2.017). This is not unexpected as
both radicals have similar electronic structures. It thus appears that the experimental
g-value of Me2NCS2• (g = 2.03) is somewhat high and may require re-examination.
The low S—S bond dissociation enthalpy is attributed to the remarkable stability of
the Me2NCS2• radical. To provide more definitive estimates of the various bond
135
dissociation enthalpies of 1, the structures and energies of species 1 to 5 were also
examined by the G3X(MP2)-RAD method, which yields more reliable predictions of
thermochemical data of free radicals [27]. From the data in Table 6.1, the computed
dissociation enthalpies obtained by both methods agree within 5 kJ mol–1. This result
is perhaps not surprising as the radicals 2, 3 and 4 do not have a significant spin
contamination in their UHF-based wave functions ( ⟩⟨2S = 0.77 in all cases). Hence,
the best theoretical estimate of the lowest dissociation enthalpy of TMTD to form two
Me2NCS2• radicals is 150.0 kJ mol–1, in perfect agreement with the value of 151 kJ
mol–1 reported by Blokh [28] who investigated the isotopic exchange between TMTD
and S8 using radioactive 35S. In this context, it is interesting to note that TMTD
triggers radical polymerization reactions of styrene and of methyl-2-methaylacrylate
already at 70°C [29].
Reaction in equation G3X(MP2) G3X(MP2)-RAD
2 155.1 150.0
3 266.4 268.0
4 111.3 117.8
6 267.0 265.2
7 111.9 115.3
Table 6.1. Calculated Reaction Enthalpies (∆H°298, kJ mol–1) Using the G3X(MP2) and G3X(MP2)-RAD Methods.
From the above data, it follows that the disproportionation or “self-saturation” of
radical 2 is thermodynamically unfavorable:
2 Me2NCS2• (2) → Me2NCS
• (3) + Me2NCS3• (4)
136
∆H°298 = 111.3 kJ mol–1 ∆G°298 = 120.5 kJ mol–1
(4)
The dissociation of one of the C—N bonds of 1 with formation of another radical (6)
has also been discussed in the literature:
(Me2NCS2)2 (1) → Me2N• (5) + Me2NC2S4
• (6)
(5)
However, radical 6 is not a stable equilibrium structure. Attempts to optimize its
geometry at the B3LYP/6-31G(2df,p) level results in a CS2 fragment dissociating
from Me2NCS2•. The overall reaction is therefore as follows and no carbon radical is
formed:
(Me2NCS2)2 (1) → Me2N• (5) + CS2 + Me2NCS2
• (2)
∆H°298 = 267.0 kJ mol–1 ∆G°298 = 173.4 kJ mol–1
(6)
Based on the calculated energies, the reported formation of a carbon radical by
thermal dissociation of TMTD at 145°C is unlikely.
The homolytic dissociation of radical 2 is less endothermic than the homolysis of the
S—S bond of TMTD:
Me2NCS2• (2) → Me2N
• (5) + CS2
137
∆H°298 = 111.9 kJ mol–1 ∆G°298 = 73.2 kJ mol–1
(7)
Therefore, at temperatures high enough to initiate the reaction in equation (2), the
reaction in equation (7) will also take place. However, since the concentration of
radical 2 will always be very small, the concentration of Me2N• will be even smaller
unless CS2 is allowed to escape, in which case Me2N• will be continuously produced.
The thermal decomposition of TMTD has been reported to yield eventually CS2,
tetramethylthiurea (TMTU) and elemental sulfur [6]. This reaction is predicted to be
slightly endothermic, but exergonic if all species are gaseous.
(Me2NCS2)2 (1) → CS2 + (Me2N)2CS + ⅛S8
∆H°298 = 2.7 kJ mol–1 ∆G°298 = −35.2 kJ mol–1
(8)
However, other studies show that the product of thermal TMTD decomposition was
not TMTU but tetramethylthiuram monosulfide (TMTM) [30]. The radicals 2, 4 and 5,
as well as tetramethylthiuram polysulfides (TMTP) are believed to be intermediates in
these reactions. In this context, the hitherto unknown structure of Me2NCS4• was
calculated to find out if it was a cyclic or chain-like species. Three structural
conformers were obtained (7, 7a and 7b) and their structures are given in Figure 6.6.
The two conformers 7 and 7a differ mainly by their torsional angles. The global
minimum structure 7 has all non-hydrogen atoms practically in one plane. Dipole
moment and NBO atomic charges are given in Table 6.2. The relative enthalpy of 7a
138
with respect to 7 is 11.3 kJ mol–1. In 7a, the sulfur atoms S3 and S4 are out of the
plane defined by the NCS2 group. Torsional angles for 7a are: N−C−S−S, −32.3°;
S−C−S−S, 149.9°; C−S−S−S, −59.1°. The third conformer 7b was obtained when a
cyclic structure with a five-membered CS4 heterocycle was constructed and geometry
optimization attempted. This species, with a relative enthalpy of 62.6 kJ mol–1 with
respect to 7, is of C2 symmetry with one rather long S—S bond of 2.700 Å connecting
the two disulfide groups. Therefore, radical 7b may be denoted as Me2NC(S2)2•. The
spin density plots of the three structures are presented in Figure 6.7. Essentially, 7 and
7a are π-type radicals with unpaired spin densities located in the terminal S—S
moiety. 7b is better described as a σ-type radical. The predicted isotropic g-values of
7, 7a and 7b are 2.024, 2.029 and 2.011 respectively. These ESR g-values may be
used to differentiate these species from the related radicals 2, 3 and 4.
139
1.348
1.662
1.796
2.148 1.957
S1
S2
S3
S4
1.342
1.660
1.789
2.165
1.960
S1
S2
S3
S4
1.342
1.740
2.040
2.700
S1
S2 S3
S4
7 (C1)
7a (C1) 7b (C2)
Figure 6.6. B3LYP/6-31G(2df,p) optimized geometries of the conformers of the Me2NCS4• radical, 7,
7a and 7b, together with the numbering convention of sulfur atoms. Distances are given in Ångstroms.
Species N S1 S2 S3 S4 µ
TMTD (1) −0.44 −0.12 +0.17 +0.17 −0.12 0.95
Me2NCS• (3) −0.45 −0.02 4.44
Me2NCS2• (2) −0.42 −0.04 −0.04 4.92
Me2NCS3• (4) −0.43 +0.06 +0.24 −0.21 6.03
Me2NCS4• (7) −0.43 −0.12 +0.15 +0.20 −0.16 5.28
Me2NCS4• (7a) −0.43 +0.06 +0.16 +0.09 −0.13 4.51
Me2NC(S2)2• (7b) −0.41 +0.26 −0.23 +0.26 −0.23 7.73
Table 6.2. NBO Atomic Charges of the Donor Atoms and Dipole Moments µ in Debye of Species 1–4, 7, 7a and 7b.
140
Figure 6.7. UB3LYP/6-31G(2df,p) spin density plots of 7, 7a and 7b.
The formation of 4 and 7 by sulfuration of 2 with elemental sulfur is slightly
endothermic and endergonic, and the corresponding equilibria lie on the left hand side
at moderate temperatures:
Me2NCS2• (2) + ⅛ S8 → Me2NCS3
• (4)
∆H°298 = 41.6 kJ mol–1 ∆G°298 = 40.1 kJ mol–1
(9)
141
Me2NCS2• (2) + ¼ S8 → Me2NCS4
• (7)
∆H°298 = 9.3 kJ mol–1 ∆G°298 = 16.8 kJ mol–1
(10)
By recombination reactions between radicals 2, 4 and 7, various tetramethylthiuram
polysides (TMTP) can be formed. Thus, one possible reaction sequence for the
formation of TMTP on heating of TMTD is 1 → 2 → 4 → 7 → TMTP. The
disproportionation of the disulfide TMTD into the monosulfide (TMTM) and the
trisulfide (TMTT) has also been discussed in the literature [31].
2 (Me2NCS)2S2 (TMTD, 1) → (Me2NCS)2S (TMTM) + (Me2NCS)2S3 (TMTT)
∆H°298 = 6.5 kJ mol–1 ∆G°298 = 9.5 kJ mol–1
(11)
(Me2NCS)2S2 (TMTD, 1) + Me2NCS2• (2) → (Me2NCS)2S (TMTM) + Me2NCS3
• (4)
∆H°298 = 47.3 kJ mol–1 ∆G°298 = 45.1 kJ mol–1
(12)
The structures of TMTU, TMTM and TMTT are shown in Figure 6.8. The trisulfide
molecule is of C2 symmetry with normal geometrical parameters. S—S bond lengths
are 2.053 Å in length, and the C−S−S−S torsional angle is 85.4°. The substituents
Me2NC(=S) are trans to each other. It is not known whether the reaction in equation
(11) proceeds via the intermediate formation of radicals, for example, as in equations
(2), (12) and (13):
142
Me2NCS2• (2) + Me2NCS3
• (4) → (Me2NCS)2S3 (TMTT)
∆H°298 = 195.9 kJ mol–1 ∆G°298 = 135.9 kJ mol–1
(13)
1.3741.677
1.341
1.6561.817
1.803
1.6621.352
1.3491.652
1.8362.053
TMTU (C2) TMTM (C1)
TMTT (C2)
Figure 6.8. B3LYP/6-31G(2df,p) optimized geometries of TMTU, TMTM and TMTT. Distances are in Ångstroms. Molecular symmetry is given in parentheses.
It is remarkable that the S—S bond dissociation enthalpy of TMTT (195.9 kJ mol–1) is
larger than that of TMTD (150.0 kJ mol –1), whereas normally disulfide bonds are
thermally more stable than the tri- and tetrasulfide chains of organic polysulfides [32,
33].
143
6.3.2 Impact of Metal Cations on the Dissociation of Tetramethylthiuram Di- and
Trisulfides.
The various radicals as well as TMTD itself contain several donor atoms with lone
electron pairs, and these are expected to interact with the positively charged metal
atoms of zinc oxide or zinc complexes added to a rubber vulcanization mixture as
essential ingredients or “activators”. The NBO atomic charges of 1 to 4 (Table 6.2)
together with the well known polarizability values indicate that the terminal sulfur
atoms are likely to be the preferred sites for any cation coordination, while the planar
coordinated nitrogen atoms are of lower Lewis basicity despite their higher negative
charge. For initial structures with a Li+ ion directly linked to the nitrogen atoms of
radicals 2 and 4, the Li+ ion moved to the sulfur atoms on geometry optimization. This
finding is in line with the structures of several metal complexes containing TMTD as
a bidentate chelate ligand linked to a metal atom by the thione sulfur atoms [34], e.g.
in [ZnI2(TMTD)] [35], and [HgI2(TMTD)] [36]. Certain organometallic complexes
react with TMTD or the corresponding ethyl derivative at 20°C with cleavage of the
S—S bond and oxidative addition of the thiocarbamato units to the metal atoms. This
holds true for [MeAuPPh3] yielding [MeAu(S2CNMe2)2] [37] and for [(CpCr(CO)3)2]
yielding [CpCr(CO)2(S2CNEt2)] [38].
To study the influence of a metal cation on the S—S bond dissociation energy of
TMTD, the smaller lithium atom was used as a probe to save computational time and
to allow for a high level of theory for theoretical investigation. All species 1 to 4 form
stable chelate complexes with a Li+ ion, which becomes two- or three-coordinated.
The optimized structures of these lithium complexes are shown in Figure 6.9.
144
2.3492.371
2.5171.681
1.683
1.323
1.8202.070
1.824
1.332
Li
1.291
2.273
1.633
2.331
Li
2.420
1.7251.310
Li 1.328
1.845
1.679
2.310
1.968
2.413
Li
8 (C1) 9 (Cs)
10 (Cs) 11 (Cs)
S1
S2
S3
S4
Figure 6.9. B3LYP/6-31G(2df,p) optimized geometries of the cationic lithium complexes [Li(TMTD)]+ (8), [Li(SCNMe2)]
•+ (9), [Li(S2CNMe2)]•+ (10) and [Li(S3CNMe2)]
•+ (11). Distances are in Ångstroms. Molecular symmetry is given in parentheses.
The structure of [Li(TMTD)]+ (8) is of C1 symmetry and characterized by a seven-
membered ring since the metal ion forms a bridge between the two thione sulfur
atoms of TMTD. The Li—S bond lengths of 2.349 and 2.371 Å are in line with the
data of other lithiated sulfur-rich compounds [39-41]. There is one additional Li…S
contact of 2.517 Å to one side of the disulfide sulfur atoms, and the metal atom may
thus be considered as three-coordinated. The structural parameters of the TMTD
ligand changed only slightly from those of the free molecule, except for the two
torsional angles S3−S2−C=S1 (134.3°) and S3−S2−C−N (−50.5°) on the left side of
the molecule 8 in Figure 6.9. A structure of C2 symmetry with two-coordinated Li+ is
145
also a minimum on the potential energy hypersurface of [Li(TMTD)]+, but this isomer
is by 10.7 kJ mol–1 less stable than 8.
The small Me2NCS• radical coordinates to a lithium cation via both the sulfur and the
carbon atom resulting in the formation of a three-membered ring. All non-hydrogen
atoms of [Li(SCNMe2)]•+ (9) are coplanar, and the molecular symmetry is Cs. The
shortest S…H contact is 2.840 Å. The structure of [Li(S2CNMe2)]•+ (10) is also of Cs
symmetry and characterized by a non-planar four-membered CS2Li heterocycle. The
metal ion is not in the plane of the C2NCS2 skeleton; the mirror plane bisects the
angles C2N and CS2. This structure is obviously a compromise between the ion-dipole
attraction, which would be maximum if Li+ was in the C2NCS2 plane and located on
the dipole vector that is in the direction of the N—C (sp2) bond, and a covalent bond
between Li+ and the sulfur 3p lone pairs, which are perpendicular to the C2NCS2 plane
and are part of the four-center π electron system of the radical. The torsional angles
S−C−S−Li are ±37.9°. The sulfur-rich radical cation [Li(S3CNMe2)]•+ (11) is also of
Cs symmetry with all non-hydrogen atoms in one plane. Both the C=S and S—S
bonds contract on Li+ coordination while the C—S bond considerably lengthens by
0.107 Å. Interestingly, the C—N bonds of 1 to 4 become shorter upon complexation
with Li+ (See Figures 6.2, 6.3 and 6.9). In general, the geometrical features of the
S…Li+ coordination in 8 to 11 are similar to those calculated for the lithium
complexes of sulfur clusters [39-41].
146
Complex ∆H°298 ∆G°298 Μ
[Li(TMTD)]+ (8) −284.9 −243.8 3.01
[Li(SCNMe2)]•+ (9) −166.1 −137.2 6.16
[Li(S2CNMe2)]•+ (10) −181.1 −149.0 4.69
[Li(S3CNMe2)]•+ (11) −253.6 −203.9 2.46
[Li(S2CNMe2)2]•+ (12) −314.3 −235.2 4.74
[Li(TMTT)]+ (14) −291.5 −254.9 3.17
[Li(S3CNMe2)2]•+ (15) −409.2 −326.0 1.42
Table 6.3. Binding energies, in kJ mol–1, and dipole moments µ, in Debye, of the lithium complexes 8–12, 14 and 15. Energies correspond to G3X(MP2) values, and dipole moments are calculated at the B3LYP/6-31G(2df,p) level.
The binding energies of the four complexes 8 to 11, that is, the energy differences
between Li+ and the free ligand on one side and the complex on the other side, are
listed in Table 6.3. These data demonstrate that the binding energies of the two-
coordinated species [Li(SnCNMe2)]•+, are smaller than that for [Li(TMTD)]+, but
these energies increase with the sulfur content of the ligand and reach a maximum for
n = 3. In the latter case, the radical cation complex contains a five-membered
metallacycle. As a consequence of the increase in binding energy of with sulfur
content, the sulfurization of 10 by elemental sulfur to produce the trithiolane
derivative 11 is exothermic and exergonic:
[Li(S2CNMe2)]•+ (10) + ⅛S8 → [Li(S3CNMe2)]
•+ (11)
∆H°298 = −19.9 kJ mol–1 ∆G°298 = −13.7 kJ mol–1
(14)
147
This finding is important, since only the polysulfidic complexes are assumed to
abstract allylic hydrogen atoms from rubber molecules with the formation of H2S.
Evidently, the polysulfido complexes are formed spontaneously if elemental sulfur is
added, in contrastto the endothermic sulfuration reactions of the free radical 2 (see
equation (9)). Using the more reactive (metastable) polymeric sulfur “Crystex” [42],
the equilibrium of the reaction in equation (14) will lie even more toward the right.
Due to the high stability of the complex [Li(TMTD)]+, the homolytic dissociation
energies of the TMTD ligand with formation of one free radical and one ligand-
stabilized radical are drastically different from those of free TMTD:
[Li(TMTD)]+ (8) → [Li(S2CNMe2)]•+ (10) + Me2NCS2
• (2)
∆H°298 = 257.6 kJ mol–1 ∆G°298 = 193.7 kJ mol–1
(15)
[Li(TMTD)]+ (8) → [Li(S3CNMe2)]•+ (11) + Me2NCS
• (3)
∆H°298 = 307.4 kJ mol–1 ∆G°298 = 245.9 kJ mol–1
(16)
[Li(TMTD)]+ (8) → [Li(SCNMe2)]•+ (9) + Me2NCS3
• (4)
∆H°298 = 384.9 kJ mol–1 ∆G°298 = 312.7 kJ mol–1
(17)
148
A more likely thermal reaction of 8 is the isomerization to the bis-chelate complex 12
with a four coordinated lithium atom and the TMTD ligand dissociated into two equal
fragments:
[Li(TMTD)]+ (8) → [Li(S2CNMe2)2]+ (12)
∆H°298 = 127.1 kJ mol–1 ∆G°298 = 153.7 kJ mol–1
(18)
The electronic ground state of species 12 is a triplet, which is by 12.0 kJ mol–1 more
stable than the singlet state. The geometry of 12 is shown in Figure 6.10. It is
interesting to note that the Li is tetrahedrally coordinated to four S atoms in this
species. The mode of binding for the Li(S2CNMe2) moieties is similar to that in 10,
where the Li+ ion binds slightly above the CS2 plane. The fact that the dissociation of
the TMTD ligand in 8 to give the dissociated ligands present in 12 requires only 127.1
kJ mol–1 compared to 155.1 kJ mol–1 for the dissociation of free TMTD at the
G3X(MP2) level, shows that the lithium cation facilitates S—S bond homolysis in
TMTD.
Li
2.506
1.717
1.314
Figure 6.10. B3LYP/6-31G(2df,p) optimized geometry of the triplet lithium complex [Li(S2CNMe2)2]+.
Distances are given in Ångstroms.
149
If additional lithium cations are available, complex 8 may be cleaved into two
identical complexes 10:
[Li(TMTD)]+ (8) + Li+ → 2 [Li(S2CNMe2)]•+ (10)
∆H°298 = 75.4 kJ mol–1 ∆G°298 = 43.6 kJ mol–1
(19)
The overall reaction of TMTD with an excess of Li+ is thus strongly exothermic:
TMTD (1) + 2 Li+ → 2 [Li(S2CNMe2)]•+ (10)
∆H°298 = −209.2 kJ mol–1 ∆G°298 = −200.0 kJ mol–1
(20)
The various dissociations of TMTD are shown schematically in Figure 6.11.
-300
-150
150
0
H°298 (kJ mol–1)
TMTD
2 Me2NCS2•
+155.1
–284.6+Li+
[Li(TMTD)]+
+127.1
–157.5+Li+
[Li(S2CNMe2)2]+
+75.4
–209.2+2 Li+
+Li+
2 [Li(S2CNMe2)]•+
Figure 6.11. Schematic energy level diagram for the homolytic dissociation and complex formation reactions of TMTD with and without Li+ cations, based on G3X(MP2) computed reaction enthalpies.
150
The reaction of 12 with S8 to the mixed ligand complex [Li(S2CNMe2)(S3CNMe2)]+
(13) in its triplet state is slightly endothermic:
[Li(S2CNMe2)2]+ (12) + ⅛ S8 → [Li(S2CNMe2)(S3CNMe2)]
+ (13)
∆H°298 = 2.2 kJ mol–1 ∆G°298 = –11.3 kJ mol–1
(21)
Attempts to optimize the structure of [Li(S2CNMe2)(S3CNMe2)]+ as a singlet resulted
in the novel species [Li(TMTT)]+ (14) in which the trisulfide ligand TMTT forms
three Li—S bonds to the metal cation. This complex is by 133.1 kJ mol–1 more stable
than 13.
TMTT + Li+ → [Li(TMTT)]+ (14)
∆H°298 = −291.5 kJ mol–1 ∆G°298 = −254.9 kJ mol–1
(22)
As a result of the high stability of 14, the following sulfurization reaction with spin-
crossover is exothermic:
[Li(S2CNMe2)2]+ (12) + ⅛ S8 → [Li(TMTT)]+ (14)
∆H°298 = −130.3 mol–1 ∆G°298 = −156.5 kJ mol–1
(23)
Dissociation of the trisulfide TMTT in the presence of 2 equivalents of Li+ is also an
exothermic process:
151
TMTT + 2 Li+ → [Li(S2CNMe2)]•+ (10) + [Li(S3CNMe2)]
•+ (11)
∆H°298 = −229.9 kJ mol–1 ∆G°298 = −218.2 kJ mol–1
(24)
The formation of the paramagnetic lithium complex [Li(S3CNMe2)2]+ (15) as shown
in equation (25), is a thermodynamically favorable process. Interestingly, the Li+ ions
of 13 and 15 are coordinated to four sulfur atoms in a tetrahedral manner.
TMTD + Li+ + ¼ S8→ [Li(S3CNMe2)2]+ (14)
∆H°298 = −187.1 kJ mol–1 ∆G°298 = −158.5 kJ mol–1
(25)
The structures of 13, 14 and 15 are shown in Figure 6.12.
152
2.377
2.490 2.528
2.530
1.717
1.717
1.315
1.963
1.671
1.849
1.331
2.707
2.508
2.318
1.818
1.681
1.332
2.387
2.507
1.963
1.669
1.8501.332
13 (C1)
14 (Cs)
15 (C2)
Li
Li
Li
Figure 6.12. B3LYP/6-31G(2df,p) optimized geometries of the triplet species 13 and 15 and singlet species 14. Distances are given in Ångstroms. Molecular symmetry is given in parentheses.
The presence of free metal cations and elemental sulfur, TMTD and the
corresponding polysulfides TMTP are predicted to spontaneously dissociate with the
formation of chelate complexes provided the activation energy is available. This
activation energy is expected to be low, since the interaction of TMTD and TMTT
with a cation starts with the exothermic coordination to thione sulfur atoms.
Zinc chloride and tetramethylthiourea form a complex [ZnCl2(TMTU)2] in which the
metal atom is coordinated by two chloride anions and the thione sulfur atoms of two
TMTU molecules in an approximately tetrahedral manner [43]. Tetracoordinated zinc
153
complexes with the related anions RCS2– [44] and RCS3
– [45, 46] have been prepared
and tetra- and hexacoordinated zinc complexes with thiocarbamato anions Me2NCS2−
[47] and Me2NCS3− [4, 5] are well known. It can reasonably be assumed that the
radicals 2–4 can function as ligands in such complexes too. Preliminary DFT
investigations using lithium ions as a probe support this view.
Three-coordinated zinc ions on the surface of bulk ZnO particles are expected to
facilitate the formation of radicals from TMTD during the vulcanization of rubber
with this accelerator system in a similar fashion as the reaction in equation (20). Since
ca. 95% of a typical vulcanization mixture consists of rather non-polar materials such
as polyolefins and carbon black, there should be no strong solvent effects. Radical
reactions may then play a major role during the vulcanization of rubber with this
particular accelerator system. The predicted paramagnetic thiocarbamato zinc
complex [Zn(S2CNMe2)]2+ is expected to be sulfurized by elemental sulfur with
formation of the analogous polysulfidic trithiolane species, which can then react when
the polyolefins with crosslinking. At higher temperatures, additional redox reactions
take place between ZnO and TMTD yielding zinc dithiocarbamate, TMTU, elemental
sulfur, CS2, and zinc sulfate [48]. Possible intermediates of this reaction have recently
been characterized [49].
6.4 Concluding Remarks
Tetramethylthiuram disulfide (TMTD) and the corresponding trisulfide (TMTT) are
predicted to dissociate on heating preferentially with homolytic dissociation of the
154
S—S bonds to give radicals. The reported g-values of some of these radicals match
those obtained experimentally from ESR spectra. The formed radicals Me2NCSn• (n =
1–4) are stabilized by electron delocalization. Interestingly, the S—S bond
dissociation enthalpy of TMTD is considerably smaller than that of TMTT, which is
rather uncommon in organosulfur chemistry. The low S—S bond dissociation
enthalpy of TMTD (150.0 kJ mol–1) is attributed to the remarkable stability of
Me2NCS2•, a σ-type radical with the unpaired spin density shared equally between
two sulfur atoms and a four-centered delocalized π system. Both the experimental
dissociation enthalpy and the ESR g-value of tetraisopropylthiuram disulfide are
reproduced well by these calculations. TMTD, TMTT and the radicals Me2NCSn• (n =
1–3) are excellent ligands toward metal cations. The high binding energies of the
complexes [Li(SnCNMe2)2]+ (n = 2, 3) are responsible for the unexpected finding that
the homolytic dissociation of the S—S bonds of both TMTD and TMTT are
facilitated by lithium cations, especially in the presence of elemental sulfur. Therefore,
spontaneous dissociation of these vulcanization accelerators is also expected in the
presence of zinc ions and consequently, also in a rubber vulcanization mixture
containing TMTD as an accelerator and zinc oxide as an activator.
References
1. Heideman, G.; Datta, R.; Noordermeer, J.; van Baarle, B. Rubber Chem.
Technol. 2004, 77, 512.
2. Heideman, G.; Datta, R. N.; Noordermeer, J. W. M.; van Baarle, B. J. Appl.
Polym. Sci. 2005, 95, 1388.
155
3. Nieuwenhuizen, P. J. Appl. Catal., A 2001, 207, 55.
4. Nieuwenhuizen, P. J.; Ehlers, A. W.; Haasnoot, J. G.; Janse, S. R.; Reedijk, J.;
Baerends, E.-J. J. Am. Chem. Soc. 1999, 121, 163.
5. Nieuwenhuizen, P. J.; Ehlers, A. W.; Hofstraat, J. W.; Janse, S. R.; Nielen, M.
W. F.; Reedijk, J.; Baerends, E.-J. Chem.--Eur. J. 1998, 4, 1816.
6. Coleman, M.; Shelton, J.; Koenig, J. Rubber Chem. Technol. 1973, 46, 957.
7. Degtyarev, L. S.; Ganyuk, L. N. Polym. Sci. U. S. S. R. 1964, 6, 31.
8. Plyusnin, V. F.; Kuznetzova, E. P.; Bogdanchikov, G. A.; Grivin, V. P.;
Kirichenko, V. N.; Larionov, S. V. J. Photochem. Photobiol., A 1992, 68, 299.
9. Nichols, P. J.; Grant, M. W. Aust. J. Chem. 1983, 36, 1379.
10. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Pople, J. A. J. Chem. Phys.
2001, 114, 108.
11. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Rassolov, V.; Pople, J. A. J.
Chem. Phys. 1999, 110, 4703.
12. Henry, D. J.; Sullivan, M. B.; Radom, L. J. Chem. Phys. 2003, 118, 4849.
13. Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. (Washington, DC, U. S.)
1988, 88, 899.
14. Messerschmidt, M.; Wagner, A.; Wong, M. W.; Luger, P. J. Am. Chem. Soc.
2002, 124, 732.
15. Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem. Phys.
1998, 108, 4439.
16. Patchkovskii, S.; Ziegler, T. J. Phys. Chem. A 2001, 105, 5490.
17. Schreckenbach, G.; Ziegler, T. J. Phys. Chem. A 1997, 101, 3388.
18. van Lenthe, E.; Wormer, P. E. S.; van der Avoird, A. J. Chem. Phys. 1997,
107, 2488.
156
19. te Velde, G.; Bickelhaupt, F. M.; Baerends, E.-J.; Fonseca Guerra, C.; van
Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. J. Comput. Chem. 2001, 22,
931.
20. Fonseca Guerra, C.; Snijders, J. G.; te Velde, G.; Baerends, E.-J. Theor. Chem.
Acc. 1998, 99, 391.
21. ADF2005.01 SCM, Theoretical Chemistry, Vriie Universiteit: Amsterdam,
The Netherlands.
22. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Montgomery, J., J. A.; Vreven, T.; Kudin, K. N.; Burant, J.
C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi,
M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara,
M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda,
Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.;
Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R.
E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala,
P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.;
Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.;
Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.;
Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.;
Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M.
W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A.
Gaussian 03, Gaussian, Inc.: Wallingford, CT, 2004.
23. Werner, H.-J.; Knowles, P. J. MOLPRO 2002, University of Birmingham:
Birmingham, 2002.
157
24. Wang, Y.; Liao, J. H. Acta Crystallogr., Sect. B: Struct. Sci. 1989, 45, 65.
25. Luo, Y.-R., Handbook of Bond Dissociation Energies in Organic Compounds.
CRC Press: Boca Raton, FL, 2003.
26. Hehre, W. J.; Radom, L.; Schleyer, P. v. R.; Pople, J. A., Ab Initio Molecular
Orbital Theory. Wiley: New York, 1986.
27. Gómez-Balderas, R.; Coote, M. L.; Henry, D. J.; Radom, L. J. Phys. Chem. A
2004, 108, 2874.
28. Blokh, G. Rubber Chem. Technol. 1958, 31, 1035.
29. Ferington, T. E.; Tobolsky, A. V. J. Am. Chem. Soc. 1955, 77, 4510.
30. Kruger, F. W. H.; McGill, W. J. J. Appl. Polym. Sci. 1991, 42, 2669.
31. Geyser, M.; McGill, W. J. J. Appl. Polym. Sci. 1995, 55, 215.
32. Benson, S. W. Chem. Rev. (Washington, DC, U. S.) 1978, 78, 23.
33. Steudel, R. Chem. Rev. (Washington, DC, U. S.) 2002, 102, 3905.
34. Victoriano, L. I. Coord. Chem. Rev. 2000, 196, 383.
35. McCleverty, J. A.; Morrison, N. J. J. Chem. Soc., Dalton Trans. 1976, 2169.
36. Beurskens, P. T.; Cras, J. A.; Noordik, J. H.; Spruijt, A. M. J. Chem.
Crystallogr. 1971, 1, 93.
37. Perevalova, E. G.; Grandberg, K. I.; Dyadchenko, V. P.; Kalinina, O. N. J.
Organomet. Chem. 1988, 352, C37.
38. Goh, L. Y.; Weng, Z.; Leong, W. K.; Leung, P. H. Angew. Chem. Int. Ed.
2001, 40, 3236.
39. Steudel, Y.; Wong, M. W.; Steudel, R. Chem.--Eur. J. 2005, 11, 1281.
40. Steudel, Y.; Wong, M. W.; Steudel, R. Eur. J. Inorg. Chem. 2005, 2005, 2514.
41. Wong, M. W.; Steudel, Y.; Steudel, R. Inorg. Chem. 2005, 44, 8908.
158
42. Steudel, R.; Passlack-Stephan, S.; Holdt, G. Z. Z. Anorg. Allg. Chem. 1984,
517, 7.
43. Kunchur, N. R.; Truter, M. R. J. Chem. Soc. 1958, 3478.
44. Bonamico, M.; Dessy, G.; Fares, V.; Scaramuzza, L. J. Chem. Soc., Dalton
Trans. 1972, 2515.
45. Bonamico, M.; Dessy, G.; Fares, V.; Scaramuzza, L. J. Chem. Soc. A 1971,
3191.
46. Fackler, J. P.; Fetchin, J. A.; Fries, D. C. J. Am. Chem. Soc. 1972, 94, 7323.
47. McCleverty, J. A.; Morrison, N. J.; Spencer, N.; Ashworth, C. C.; Bailey, N.
A.; Johnson, M. R.; Smith, J. M. A.; Tabbiner, B. A.; Taylor, C. R. J. Chem.
Soc., Dalton Trans. 1980, 1945.
48. Nieuwenhuizen, P. J.; Van Duin, M.; Haasnoot, J. G.; Reedijk, J.; McGill, W.
J. J. Appl. Polym. Sci. 1999, 73, 1247.
49. Brayton, D. F.; Tanabe, K.; Khiterer, M.; Kolahi, K.; Ziller, J.; Graeves, J.;
Farmer, P. J. Inorg. Chem. 2006, 45, 6064.
159
Chapter Seven
Dissociation Enthalpies and Activation Barriers of Various
Bis(thiocarbonyl)disulfides R−C(=S)−S−S−C(=S)−R
7.1 Introduction
The curing or vulcanization of natural rubber remains the essential crux of rubber
processing. Rubber vulcanization essentially involves the cross-linking of individual
rubber polymer chains (cis-polyisoprene) via sulfide bridges, making the overall
material tougher and resistant to chemical attack. The discovery of vulcanization
using elemental sulfur has been credited to Charles Goodyear, but it has recently been
discovered that Mesoamericans have achieved this method of rubber vulcanization as
early as 1600BC [1]. In this present day, a vast number of reagents associated with
sulfur vulcanization of natural rubber have been developed. These include
dithiocarbamates, thiurams, thiourea, thiazoles, and sulfenamides [2]. Examples of
these reagents are shown in Figure 7.1.
160
Dithiocarbamates
e.g. Zinc Dimethyldithiocarbamate
(ZDMC)
Thiurams
e.g. Tetramethylthiuram Disulfide
(TMTD)
Thioureas Thiazoles
e.g. Ethylene Thiourea
(ETU)
e.g. Mercaptobenzathiazole Disulfide
(MBTS)
Sulfenamides
e.g. N-cyclohexyl-2-benzothiazole sulfenamide
(CBS)
Dithiocarbamates
e.g. Zinc Dimethyldithiocarbamate
(ZDMC)
Dithiocarbamates
e.g. Zinc Dimethyldithiocarbamate
(ZDMC)
Thiurams
e.g. Tetramethylthiuram Disulfide
(TMTD)
Thiurams
e.g. Tetramethylthiuram Disulfide
(TMTD)
Thioureas Thiazoles
e.g. Ethylene Thiourea
(ETU)
e.g. Mercaptobenzathiazole Disulfide
(MBTS)
Sulfenamides
e.g. N-cyclohexyl-2-benzothiazole sulfenamide
(CBS)
Figure 7.1. Examples of different sulfur vulcanization reagents commercially available.
Previous work on the accelerator tetramethylthiuram disulfide (TMTD) showed that
its S—S bond dissociation enthalpy (150 kJ mol–1) was lower than the corresponding
trisulfide (TMTT) (190 kJ mol–1) based on high level calculations using the
G3X(MP2) composite method [3]. The dependence of the dissociation enthalpies of
the bis(thiocarbonyl)disulfides R−C(=S)−S−S−C(=S)−R, with R = H, Me, F, Cl, OMe,
SMe, NMe2 and PMe2 is investigated in this work. The better known
bis(thiocarbonyl)disulfides are those with R = Alkyl/Aryl [4], OMe [5], SMe [6, 7]
and NMe2 [8]. Less is known about the chlorine substituted case (R = Cl) [9], and no
references have been found for R = F and PMe2. Experimental structure
determinations of these compounds have not been published to date. However, limited
theoretical structures have been reported for R = H (at the HF/STO-3G level) [4] and
OMe (at the CNDO level) [10]. An important goal of this work is to obtain structural
and thermochemical data for the eight compounds mentioned above at a high level of
theory for the first time.
161
Many coordination compounds are known which contain the above compounds as
ligands or anionic ligands derived from these species by reductive cleavage of the S—
S bond [11]. To systematically investigate the ligand properties of the neutral species
(RCS2)2 and the radicals RCS2• that form as a result of S—S bond homolysis, with R
= H, Me, F, Cl, OMe, SMe, NMe2 and PMe2, the lithium cation is used as a model
metal cation for ligand coordination. The small Li+ cation allows for a high level of
theory to be used in quantum-chemical calculations. Since the binding energies of Li+
complexes towards a large number of Lewis bases are linearly related to those of
other cations such as Na+ and Mg2+ [12], the use of Li+ as a model cation seems
justified.
As mentioned in previous work, the stability of the Me2NCS2• radical is remarkable,
as a result of effective delocalization of the π orbitals and the spin density over the
four-center NCS2 framework [3]. However, the S—S bond dissociation enthalpy
(BDE) for TMTD is still endothermic by 155.1 kJ mol–1 at the G3X(MP2) level [3].
In industrial rubber vulcanization, activators such as zinc oxide are added to facilitate
S—S bond breaking by the coordination of TMTD to zinc. The binding of small
sulfur containing compounds to zinc oxide clusters was investigated by Steudel et al,
and it was found that the binding energy Zn4O4 clusters to S6, Me2S2, MeSH and
Me2S was exothermic by –60.7 to –106.7 kJ mol–1 [13]. Zinc oxide, however, is an
environmental pollutant that displays eco-toxicity [14] and the inhalation of zinc
oxide can result in severe pulmonary conditions [15, 16]. In an effort to find ways that
can further stabilize the RCS2• radical and hence reduce the reliance on zinc oxide
activators, suitable substituents R were considered based on their electron donating
162
ability as well as their ability to form resonance stabilized structures. The electron
donating ability of the –OSF5 group is demonstrated by its strong conjugate acid
HOSF5 [17], and the other two guanidine substituents (hereby denoted Gu1 and Gu2)
(See Figure 7.2) are noted for their strong basicity [18] as well as ability to form
numerous resonance-stabilized structures that result in the stabilization of the RCS2•
radical. Thus, bis(thiocarbonyl)disulfides with these chosen novel substituents were
also examined theoretically.
(Gu1) (Gu2)
Figure 7.2. The homolytic dissociation of (RCS2)2 with novel substituents R.
7.2 Computational Methods
The structures and energies of the disulfides (RCS2)2, radicals RCS2•, and their
lithiated complexes for R = H, Me, F, Cl, OMe, SMe, NMe2 and PMe2 were examined
at the G3X(MP2) level of theory [19]. The G3X(MP2) method corresponds
effectively to energies calculated at the QCISD(T)/G3XL//B3LYP/6-31G(2df,p) level,
with certain assumptions made about additivity [20]. Correction terms considered
include the spin-orbit correction term, and a high-level empirical correction term to
address remaining deficiencies. Three important changes in G3X(MP2) theory from
G3(MP2) theory are important for the proper description of the molecules studied.
163
First, the use of B3LYP/6-31G(2df,p) geometry, second, the use of B3LYP/6-
31G(2df,p) zero-point vibrational energy scaled by an empirical factor of 0.9854, and
third, an addition of a g polarization function to the G3Large basis set for second-row
atoms at the Hartree-Fock level.
Structures of the novel bis(thiocarbonyl)disulfides (F5SOCS2)2, ((Gu1)CS2)2 and
(Gu2)CS2)2 and their associated radicals F5SOCS2•, (Gu1)CS2
• and (Gu2)CS2• were
optimized at the B3LYP/6-31G(d) level. Singlet biradical transition states
corresponding to homolytic S—S bond cleavage were optimized at this level of theory
as well, using the UHF formalism and the mixing of the HOMO and LUMO to
destroy α and β spatial symmetries. Energy calculations were then carried out at the
MP2(fc)/6-311+G(d,p) level of theory based on the B3LYP/6-31G(d) optimized
structures.
The RHF formalism was employed for all closed shell species, and the UHF
formalism was employed for all open-shell and singlet diradical species. Frequency
calculations were carried out to determine the nature of minima calculated, those with
all real frequencies correspond to local minima, and those with one and only one
imaginary frequency correspond to transition state structures. All calculations were
carried out using the GAUSSIAN 03 [21] and MOLPRO 2002 [22] suite of programs.
7.3 Results and Discussion
The typical structures of (RCS2)2, RCS2•, [Li(RCS2)2]
+, [Li(RCS2)]•+ and triplet
164
[(RCS2)Li(S2CR)]+ together with their atom numbering conventions, are shown in
Figure 7.3.
(RCS2)2
(C2)RCS
2•
(C2v or Cs)
[Li(RCS2)2]+
(C1)
[Li(RCS2)]•+
(Cs or C1)[(RCS2)Li(S2CR)]
+
(C2)
Figure 7.3. Typical structures of (RCS2)2, RCS2
•, [Li(RCS2)2]+, [Li(RCS2)]
•+ and triplet [(RCS2)Li(S2CR)]+. Molecular symmetry is indicated in parentheses.
7.3.1 Substituted bis(thiocarbonyl)disulfides, R = H, Me, F, Cl, OMe, SMe, NMe2,
PMe2.
Selected calculated structural parameters of (RCS2)2 for R = H, Me, F, Cl, OMe, SMe,
NMe2, PMe2 are presented in Table 7.1. Consistent with previous work done for
tetramethylthiuram disulfide (TMTD) [3], the most stable conformers of substituted
(RCS2)2 are of C2 symmetry. C=S bond lengths range from 1.603 Å for the fluorine
substituent, to 1.650 Å for the dimethylamino substituent. S—S bond lengths range
from 2.023 Å for the dimethylamino substituent, to 2.065 Å for the parent compound
(R = H). The C—R bond length interestingly shows the most variation for the various
165
substituents. For R = H, Me, F and Cl, the calculated C—R distances are slightly
shorter than the average experimental value. This trend is reversed for R = OMe, SMe,
NMe2 and PMe2. The biggest changes occur for R = OMe and NMe2 (difference in
bond lengths of 0.076 and 0.081 Å, respectively), which is indicative of the
delocalization of π electrons in the OCS2 and NCS2 frameworks. This extent of
delocalization becomes less pronounced for the third row analogues R = SMe and
PMe2, with the difference in bond lengths at 0.048 and 0.014 Å, respectively. It is also
interesting to note that in the case of the structure of (Me2PCS2)2, the C2PCS2 skeleton
is not planar but pyramidal at the phosphorus atom, unlike TMTD which has an
almost planar C2NCS2 framework (See Figure 7.4). This is not unusual, as phosphorus
prefers non-hybridization and a high s-orbital character of the lone-pair electrons, and
hence shows a reluctance to undergo s,p-hybridization to form planar coordinated
compounds [23, 24].
Substituent C1=S2
(= C5=S6) S3—S4
C1—R7 (= C5—R8)
Average Expt’l C—R
H 1.614 2.065 1.095 1.092 (C—H) Me 1.622 2.049 1.512 1.501 (C—C) F 1.603 2.045 1.331 1.327 (C—F) Cl 1.604 2.058 1.767 1.752 (C—Cl) OMe 1.628 2.034 1.327 1.405 (C—O) SMe 1.626 2.037 1.755 1.803 (C—S) NMe2 1.650 2.023 1.347 1.428 (C—N) PMe2 1.631 2.059 1.844 1.858 (C—P) Table 7.1. Selected calculated bond lengths (in Ångstroms) of (RCS2)2. Molecular geometry was optimized at the B3LYP/6-31G(2df,p) level of theory. Average experimental values for C—R bond lengths are included for comparison.
166
C5
S3S2
C2
C6
C4 P2
C3
S4
P1
S1
C1
Figure 7.4. The B3LYP/6-31G(2df,p) optimized structure of (Me2PCS2)2, showing the lack of planarity at the C2PCS2 framework.
The radicals RCS2• are typically of C2v (R = H, F, Cl, NMe2) or Cs symmetry (R = Me,
OMe, SMe, PMe2). Selected calculated structural parameters are shown in Table 7.2.
The RCS2• radical is resonance stabilized, as evidenced from the C—S bond lengths
of 1.659 to 1.700 Å, intermediate between the average length of a C—S single bond
(1.803 Å) and a C=S double bond (1.584 Å). The C—R bonds are also slightly shorter
than those in the corresponding dimers (RCS2)2, suggesting some increased extent of
delocalization of π electrons into the attachment atom of the R group (i.e. C for Me, O
for OMe, N for NMe2) (See Figure 6.5 in Chapter 6). The spin density plots of the
RCS2• radicals (see Figure 7.5) show that they are essentially σ-type, with the
unpaired spin density located in the two sulfur atoms.
167
Substituent C1—R4 C1—S H 1.089 1.665 Me 1.495 1.674 F 1.315 1.659 Cl 1.740 1.659 OMe 1.315 1.683a
1.679b SMe 1.728 1.685c
1.679d NMe2 1.335 1.700 PMe2 1.816 1.683
Table 7.2. Selected bond lengths (in Ångstroms) of RCS2
•. Molecular geometry was optimized at the B3LYP/6-31G(2df,p) level of theory. atrans- with respect to O–CH3.
bcis- with respect to O–CH3.
ctrans- with respect to S–CH3.
dcis- with respect to S–CH3.
Figure 7.5. UB3LYP/6-31G(2df,p) spin density plots of the radicals RCS2•.
168
∆H°298 and ∆G°298 for the homolytic S—S bond dissociation reaction (RCS2)2 → 2
RCS2• calculated at the G3X(MP2) level of theory are tabulated in Table 7.3. From
these results, TMTD has the lowest homolytic S—S bond dissociation enthalpy (BDE)
by far, attributed to the significant stability of Me2NCS2•, while the other substituted
(RCS2)2 have homolytic S—S BDEs in excess of 190 kJ mol–1.
(RCS2)2 → 2 RCS2
• Substituent ∆H°298 (kJ mol–1) ∆G°298 (kJ mol–1) H 210.8 158.4 Me 193.0 139.1 F 190.0 137.1 Cl 215.2 161.9 OMe 206.5 153.1 SMe 195.7 146.5 NMe2 155.1 100.2 PMe2 198.9 141.7
Table 7.3. ∆H°298 and ∆G°298 for the homolytic S—S bond dissociation reaction (RCS2)2 → 2 RCS2
• calculated at the G3X(MP2) level of theory.
An NBO charge analysis of (RCS2)2 and RCS2• was carried out at the B3LYP/6-
31G(2df,p) level, results of which are presented in Table 7.4. The NBO charge on S3
and S4 for (RCS2)2 is generally above 0.200, except for TMTD (R = NMe2) showing
a less positive charge of 0.172. For the RCS2• radical, the charges on S2, S3, S4 and
S6 even out, as compared to a more polarized situation in (RCS2)2. Along with the
two C1—S bonds effectively reaching the same bond length, this spreading of
electron density as seen in the NBO charges in C1, S2 and S3 is indicative of
resonance stabilization in the RCS2• radical. For the RCS2
• radical, charges on the R
groups for R = F, OMe and NMe2 are negative, possibly due to the electron
withdrawing effect of the electronegative attachment atoms F, O and N respectively.
169
NBO Charges R7 (=R8)a C1 (=C5) S2 (=S6) S3 (=S4) (RCS2)2 R = H 0.263 –0.608 0.090 0.255 Me 0.068 –0.373 0.069 0.236 F –0.292 –0.012 0.078 0.226 Cl 0.063 –0.503 0.164 0.277 OMe –0.135 –0.038 –0.040 0.212 SMe 0.352 –0.626 0.036 0.237 NMe2 0.068 –0.126 –0.115 0.172 PMe2 0.455 –0.742 0.044 0.242 RCS2
• R = H 0.255 –0.606 0.176 0.176 Me 0.071 –0.392 0.160 0.160 F –0.283 –0.044 0.164 0.164 Cl 0.087 –0.547 0.230 0.230 OMe –0.120 –0.075 0.057 0.137 SMe 0.385 –0.671 0.123 0.164 NMe2 –0.175 –0.165 0.040 0.040 PMe2 0.480 –0.770 0.145 0.145 Table 7.4. NBO charges of (RCS2)2 and RCS2
•, evaluated at the B3LYP/6-31G(2df,p) level of theory, for substituents R = H, Me, F, Cl, OMe, SMe, NMe2 and PMe2.
aRepresents the sum of NBO charges on the atomic centers in substituent R.
The effect of metal ion coordination on the S—S BDEs was investigated, using the
Li+ ion as a model metal cation. As with TMTD, the preferred Li+ coordination sites
for (RCS2)2 were actually at the terminal sulfur atoms instead of the attachment atom
of the substituent. Coordination of (RCS2)2 to Li+ ions generally leads to (RCS2)2
forming three favorable Li+…S interactions with a reduction in molecular symmetry
from C2 to C1 (see Figure 7.3). Selected calculated geometrical parameters of the
[Li(RCS2)2]+ complexes are presented in Table 7.5. All three Li+…S interactions for
all eight complexes studied are below 2.8 Å in length, and are, thus, expected to be
bonding interactions with a electron density maximum along their path [25]. For the
radicals [Li(RCS2)]•+, the Li+ ion is slightly out of the plane defined by C1–S2–S3
170
(See Figure 7.3). As already mentioned in previous work with the Me2NCS2• radical,
Li+ is at a compromised position between two forces: the ion-dipole attraction which
is maximum on the CS2 plane, and a covalent bond interaction with the sulfur 3p lone
pairs perpendicular to the CS2 plane. This positioning of the Li+ ion is also seen for
the triplet [(RCS2)Li(S2CR)]+ complexes. The Li+ is coordinated to two RCS2•
radicals via their terminal sulfur atoms, and in a tetrahedral manner. Overall C2
molecular symmetry is observed for these complexes. Selected geometrical
parameters for the [Li(RCS2)]•+ and triplet [(RCS2)Li(S2CR)]+ complexes are shown
in Table 7.6 and 7.7, respectively.
Substituent Li—S2 Li—S3 Li—S5 S3—S4 C1—R7 C6—R8 H 2.510 2.541 2.456 2.087 1.091 1.091 Me 2.474 2.483 2.433 2.080 1.489 1.508 F 2.506 2.561 2.451 2.073 1.296 1.308 Cl 2.494 2.515 2.456 2.076 1.701 1.722 OMe 2.452 2.496 2.410 2.062 1.291 1.301 SMe 2.441 2.482 2.409 2.064 1.704 1.725 NMe2 2.371 2.517 2.349 2.069 1.323 1.332 PMe2 2.423 2.476 2.399 2.086 1.791 1.824 Table 7.5. Selected calculated bond lengths (in Ångstroms) of [Li(RCS2)2]
+. Molecular geometry was optimized at the B3LYP/6-31G(2df,p) level of theory.
Substituent Li—S2 Li—S3 C1—S2 C1—S3 C1—R7 H 2.521 2.521 1.675 1.675 1.089 Me 2.479 2.479 1.690 1.690 1.484 F 2.513 2.513 1.679 1.679 1.287 Cl 2.504 2.504 1.684 1.684 1.689 OMe 2.463 2.450 1.704 1.708 1.282 SMe 2.456 2.439 1.705 1.713 1.691 NMe2 2.419 2.419 1.725 1.725 1.310 PMe2 2.431 2.431 1.712 1.712 1.771 Table 7.6. Selected calculated bond lengths (in Ångstroms) of [Li(RCS2)]
•+. Molecular geometry was optimized at the B3LYP/6-31G(2df,p) level of theory.
171
Substituent Li—S2
(=Li—S5) Li—S3
(=Li—S4) C1—S2
(=C6—S5) C1—S3
(=C6—S4) C1—R7
(=C6—R8) H 2.585 2.584 1.671 1.672 1.089 Me 2.551 2.550 1.685 1.685 1.486 F 2.567 2.568 1.675 1.675 1.290 Cl 2.565 2.564 1.678 1.679 1.695 OMe 2.516 2.539 1.701 1.698 1.288 SMe 2.513 2.541 1.706 1.698 1.697 NMe2 2.505 2.506 1.717 1.718 1.314 PMe2 2.520 2.523 1.702 1.702 1.782 Table 7.7. Selected calculated bond lengths (in Ångstroms) of triplet [(RCS2)Li(S2CR)]+. Molecular geometry was optimized at the B3LYP/6-31G(2df,p) level of theory.
Coordination of (RCS2)2 and RCS2• to Li+ is usually exothermic and exergonic, as
shown in Table 7.8. The case for R = NMe2 (TMTD) has the highest binding energy
of the bis(thiocarbonyl)disulfide and the disulfide radical to Li+, while the case for R
= F has the lowest binding energy. The low binding energy for (FCS2)2 and FCS2• can
be explained by the highly electronegative fluorine atom drawing electron density
away from the CS2 moiety, resulting in less pronounced ion-dipole attraction. This is
evidenced by NBO charges on fluorine of –0.292 for (FCS2)2 and –0.283 for FCS2•;
the NBO charge on the sulfur atoms of (FCS2)2 are 0.078 and 0.226, and those on
FCS2• are 0.164. Although the attachment atoms for for R = OMe and NMe2 are also
highly electronegative, the contribution of electron density from the methyl groups to
the oxygen and nitrogen atoms leads to effective back-donation of electron density
from oxygen and nitrogen lone pairs to the CS2 moiety, resulting in less positive NBO
charges on the sulfur atoms and more favorable Li+…S interaction. The NBO charges
on sulfur atoms for the cases R = OMe and NMe2 are: for (MeOCS2)2, –0.040 and
0.212, for MeOCS2•, 0.057 and 0.137, for (Me2NCS2)2, –0.115 and 0.172, and for
Me2NCS2•, 0.040 (See Table 7.4).
172
Li+ + (RCS2)2 → [Li(RCS2)2]
+ Li+ + (RCS2)• → [Li(RCS2)]
•+ Substituent ∆H°298 (kJ mol–
1) ∆G°298 (kJ mol–
1) ∆H°298 (kJ mol–
1) ∆G°298 (kJ mol–
1) H –166.3 –131.4 –101.6 –72.1 Me –199.3 –163.7 –122.7 –93.7 F –137.6 –103.2 –80.6 –50.6 Cl –152.6 –117.5 –101.5 –71.3 OMe –219.1 –184.4 –138.9 –108.7 SMe –227.0 –188.9 –142.6 –111.8 NMe2 –284.9 –243.8 –181.1 –149.0 PMe2 –223.6 –192.0 –139.9 –109.9 Table 7.8. ∆H°298 and ∆G°298 for the coordination of (RCS2)2 and RCS2
• to Li+ in kJ mol–1, calculated at the G3X(MP2) level of theory.
Subsequent Li+ mediated S—S homolytic bond scission in (RCS2)2 can occur via a
number of modes. These can also be facilitated by additional Li+ ions or the
incorporation of elemental sulfur to form bis(thiocarbonyl) oligosulfides, followed by
formation of disulfidic, trisulfidic, and polysulfidic radicals. The following three
modes are considered in this study:
[Li(RCS2)2]+ → [Li(RCS2)]
•+ + (RCS2)• (7-1)
[Li(RCS2)2]+ → [(RCS2)Li(S2CR)] + (7-2)
[Li(RCS2)2]+ + Li+ → 2 [Li(RCS2)]
•+ (7-3)
Gas-phase enthalpies and free energies of the above reactions are generally
endothermic, as shown in Table 7.9. S—S BDEs for reaction (7-1) are highly
endothermic by more than 200 kJ mol–1 for all substituents. The fluorine substituent
shows the lowest enthalpy at 247.0 kJ mol–1 and the highest is that for the methoxy
substituent, at 286.7 kJ mol–1. The enthalpies for reaction (7-1) show little variation
across different R, and are higher than those for reaction (7-2) and (7-3). The lower
enthalpies for reaction (7-2) compared to reaction (7-1) show that it is more favorable
173
for S—S bond dissociation with associated curve crossing to occur while the (RCS2)2
ligand is still chelated to the Li+ ion. Another explanation for the lower enthalpy is the
formation of an additional Li+…S contact which provides some energy for dissociation
of the S—S bond. Trends for the S—S BDE for reaction (7-2) are not the same for
those for reaction (7-1). The highest S—S BDE for reaction (7-3) is that for the parent
compound (R = H) at 198.3 kJ mol–1, and the lowest is that of TMTD (R = NMe2) at
127.0 kJ mol–1. The addition of Li+ ions aid the process of S—S bond dissociation
further, shown in the enthalpies for reaction (7-3). The enthalpy trend for reaction (7-
3) is similar to that for reaction (7-2), again with the parent compound having the
highest reaction enthalpy of 173.9 kJ mol–1 and TMTD having the lowest reaction
enthalpy of 78.2 kJ mol–1.
[Li(RCS2)2]+
→ [Li(RCS2)]•+ +
(RCS2)•
[Li(RCS2)2]+
→ [(RCS2)Li(S2CR)]+ [Li(RCS2)2]
+ + Li+ → 2 [Li(RCS2)]
•+
Substituent ∆H°298 (kJ mol–1)
∆G°298 (kJ mol–1)
∆H°298 (kJ mol–1)
∆G°298 (kJ mol–1)
∆H°298 (kJ mol–1)
∆G°298 (kJ mol–1)
H 275.5 217.7 198.3 174.2 173.9 145.6 Me 269.6 209.2 176.7 151.1 146.9 115.5 F 247.0 189.7 183.7 161.7 166.4 139.1 Cl 266.2 208.0 187.3 164.1 164.6 136.6 OMe 286.7 228.9 180.2 157.3 147.9 120.2 SMe 280.2 223.7 175.2 154.6 137.6 111.9 NMe2 259.3 195.4 127.0 100.3 78.2 46.4 PMe2 282.5 223.8 173.9 151.2 142.6 113.9 Table 7.9. ∆H°298 and ∆G°298 for the reactions [Li(RCS2)2]
+ → [Li(RCS2)]•+ + (RCS2)
•, [Li(RCS2)2]+ →
[(RCS2)Li(S2CR)]+, and [Li(RCS2)2]+ + Li+ → 2 [Li(RCS2)]
•+ in kJ mol–1, calculated at the G3X(MP2) level of theory.
Breaking of the S—S bonds in bis(thiocarbonyl)disulfides seem to be unfavorable
processes, with S—S BDEs being generally endothermic by more than 100 kJ mol–1.
The addition of Li+ ions help facilitate this process by forming energetically favorable
Li+…S interactions. Taking the enthalpy values from Tables 7.8 and 7.9, Li+-mediated
S—S bond dissociation via reaction (7-3) is most favorable for TMTD (R = NMe2)
174
with ∆H°298 and ∆G°298 values being highly exothermic and exergonic, by –206.7 and
–197.5 kJ mol–1, respectively.
7.3.2. Novel bis(thiocarbonyl)disulfides, R = OSF5, Gu1, Gu2
As discussed in the previous section, the substituted bis(thiocarbonyl)disulfides
(RCS2)2 (R = H, CH3, F, Cl, OMe, SMe, and PMe2) display more endothermic S—S
BDE than TMTD, and thus, there remains a reliance on a metal cation to facilitate
S—S homolysis. In an attempt to find suitable substituents R than can lower this S—S
BDE, three novel bis(thiocarbonyl)disulfides with R = OSF5, Gu1 and Gu2 are
explored and evaluated. B3LYP/6-31G(d) optimized structures of these novel
bis(thiocarbonyl)disulfides are shown, together with selected geometrical parameters,
in Figure 7.6. Optimized structures of their associated radicals RCS2• are shown in
Figure 7.7.
The three novel bis(thiocarbonyl)disulfides studied generally display C2 molecular
symmetry. S—S bond lengths for all three cases are generally greater than 2.0 Å, and
it is observed that the S—S bond length for ((Gu1)CS2)2 is markedly longer than the
others, at 2.111 Å. From the molecular structure in Figure 7.6, it can be seen that the 2
Gu1 groups are relatively far apart, thereby ruling out steric effect as a reason for the
relatively long S—S bond. The C—O bond length in (F5SOCS2)2, and the C—N bond
lengths in ((Gu1)CS2)2 and (Gu2)CS2)2 are calculated to be 1.367, 1.379 and 1.380 Å,
respectively. Similar to the case for R = OMe and NMe2, these bond lengths are
shorter than the average experimental C—O and C—N single bond lengths (see Table
7.1), which suggests that there is some extent of π back-donation from O and N back
175
to the CS2 moiety, leading to increased C—O and C—N bond order and hence shorter
bond lengths. The C—O and C—N bond lengths of the RCS2• radicals show little
change with respect to their corresponding dilsulfides. It is interesting to note, that the
structure of the (Gu1)CS2• radical shows a noticeable difference in the C1—S1 and
C1—S2 bonds (See Figure 7.7), which suggests a lesser degree of delocalization of
electron density in the CS2 moiety.
S2S3
F2
O1
F6
C1
C2O2
F5
S6
F1
F9
S4S5
F3
S1
F7
F4
F8
F10
C11
C12
C7
N6S1S4N3
C13
C8
C10
C2
C3
S3
N5
C1
N1
N4
C9
N2
C4
S2
C14
C5
C16
C6
C15
C4
C5
S2S3
N1
C7
N3
C1
C8 C2
C11
C3
N2
S4
S1
N5
N4
C6
C10
C9
C13N6
C14
C12
1.6131.367
1.8302.047
(F5SOCS2)2
1.6641.379
1.827 2.111
((Gu1)CS2)2
1.380
1.627
1.867 2.033
((Gu2)CS2)2
Figure 7.6. B3LYP/6-31G(d) optimized structures of (F5SOCS2)2, ((Gu1)CS2)2 and ((Gu2)CS2)2. Distances are in Ångstroms.
176
S1
C1
F2
O1
S2
F5
S3
F1
F3
F4
C4
S1C3
C1N1
S2
C5
C2
N3
N2
C6
C8
C7
S1
C3
C4
C1N1
N2
S2
C2
C6
N3
C5
C7
F5SOCS2• (Gu1)CS2
•
(Gu2)CS2•
1.356 1.682
1.671
1.390 1.684
1.717
1.3721.685
1.702
Figure 7.7. B3LYP/6-31G(d) optimized structures of F5SOCS2•, (Gu1)CS2
• and (Gu2)CS2•. Distances
are in Ångstroms.
MP2/6-311+G(d,p)//B3LYP/6-31G(d) zero-point corrected energies (∆E0) for the
homolytic S—S bond dissociation in these novel bis(thiocarbonyl)disulfides are
presented in Table 7.10. Calculated ∆E0 using this method for R = NMe2 and
G3X(MP2) ∆E0 are presented alongside for purposes of comparison.
(RCS2)2 → 2 RCS2
• Substituent Method ∆E0 (kJ mol–1) NMe2 G3X(MP2) 155.6 NMe2 MP2/B3LYPa 149.2 OSF5 MP2/B3LYPa 188.9 Gu1 MP2/B3LYPa 115.9 Gu2 MP2/B3LYPa 151.4
Table 7.10. Zero-point corrected reaction energies (∆E0) for the reaction (RCS2)2 → 2 RCS2
• , in kJ mol–1, for R = NMe2, OSF5, Gu1 and Gu2. aMP2/6-311+G(d,p)//B3LYP/6-31G(d) + ZPE.
177
For the TMTD case (R = NMe2), the relatively good agreement of ∆E0 obtained at the
MP2/6-311+G(d,p)//B3LYP/6-31G(d) level of theory with G3X(MP2) results
(difference of 6.4 kJ mol–1 or about 1.5 kcal mol–1) gives the confidence that this is an
adequate method of choice for studying the energetics of these
bis(thiocarbonyl)disulfide systems. As seen from Table 7.10, the cases for R = OSF5
and Gu2 are more endothermic than in the TMTD case, by 39.7 and 2.2 kJ mol–1,
respectively. However, the homolytic S—S dissociation reaction ((Gu1)CS2)2 → 2
(Gu1)CS2• is less endothermic than that for TMTD, by 33.3 kJ mol–1! This finding
suggests that ((Gu1)CS2)2 may be a viable alternative to TMTD as a vulcanization
reagent, as (Gu1)CS2• radicals can be formed more easily. Since S—S BDE for R =
NMe2 is lower than those for the other substituents studied, that is R = H, Me, F, Cl,
OMe, SMe, and PMe2, it can be concluded that the S—S BDE for the case R = Gu1 is
expected to be less endothermic than those for the other substituents.
A glance at the S—S dissociation barriers would shed some more light on the
energetics of this reaction for various substituents R. The reaction (RCS2)2 → 2 RCS2•
is expected to proceed via a singlet diradical transition state near the point where the
singlet potential energy hypersurface (PES) of (RCS2)2 crosses with the triplet PES of
2 RCS2•. Schematic potential energy surfaces of homolytic S—S dissociation reaction
in (RCS2)2 for R = NMe2, OSF5, Gu1 and Gu2 are shown in Figure 7.8.
178
-50
0
50
100
150
200
250
300
Reaction Coordinate
Energy (kJ mol–1)
(RCS2)2
[RCS2•…•S2CR]‡
2 RCS2•
(0.0)
(193.0)
(211.2)
(213.3)
(268.5)
(115.9)
(149.2)
(151.4)
(188.9)
Figure 7.8. Schematic potential energy profile of the homolytic S—S bond dissociation reaction (RCS2)2 → 2 RCS2
•, for R = NMe2, OSF5, Gu1 and Gu2, calculated at the MP2/6-311+G(d,p)//B3LYP/6-31G(d) + ZPE level of theory. Relative energies are given in parentheses.
The activation barrier for homolytic S—S bond dissociation reactions as shown in
Figure 7.8 is about 200 kJ mol–1. In the case for R = OSF5, the barrier is significantly
higher than that for R = NMe2, Gu1 and Gu2, at 268.5 kJ mol–1. The barrier for R =
Gu2 is slightly higher than that for TMTD, by 2.1 kJ mol–1. These two novel
bis(thiocarbonyl)disulfides are still less effective than TMTD in providing RCS2•
radicals without the aid of metal ions or elemental sulfur. However, the barrier for R =
Gu1 is calculated to be 193.0 kJ mol–1, which is 18.2 kJ mol–1 lower than that for S—
S bond dissociation in TMTD. On the basis of these results, ((Gu1)CS2)2 is predicted
to be more effective than TMTD in providing RCS2• radicals in the process of rubber
vulcanization. As the reaction energy for ((Gu1)CS2)2 → 2 (Gu1)CS2• is still
179
endothermic by about 115.9 kJ mol–1, there still would be some reliance on activators
like ZnO to drive the reaction forward. In summary, ((Gu1)CS2)2 is predicted to be a
potential reagent for sulfur vulcanization, displaying energetics which are both
kinetically and thermodynamically more favorable than TMTD. This finding also
provides an avenue to explore the use of highly electron donating substituents R, for
instance, polyguanidine substituted bis(thiocarbonyl)disulfides as viable reagents in
rubber vulcanization. Such efforts would be limited by the size of the substituents
themselves, as a bulky group could hinder the cross-linking process instead of
promoting it.
7.4 Concluding Remarks
The structures and energetics of substituted bis(thiocarbonyl)disulfides (RCS2)2, their
associated thiocarbamate radicals RCS2•, and their coordinated compounds with
lithium cation, for R = H, Me, F, Cl, OMe, SMe, NMe2 and PMe2 have been studied
theoretically, to determine their usefulness as reagents in the sulfur vulcanization of
rubber. It was found that S—S BDEs for these (RCS2)2 are about 200 kJ mol–1, at the
G3X(MP2) level, with the exception of TMTD (R = NMe2) having the least
endothermic S—S BDE at 155.1 kJ mol–1. The addition of Li+ ions to (RCS2)2
promotes the homolytic S—S bond dissociation process by providing energy from
the exothermic association of Li+ to (RCS2)2. This process is likely to proceed via the
formation of a triplet complex [(RCS2)Li(S2CR)]+; the two RCS2• radicals are bound
to the Li+ ion via their sulfur atoms and in a tetrahedral manner. Additional Li+ ions
would facilitate the process by forming [Li(S2CR)]•+ radicals, and for the TMTD case,
this process is by far the most exothermic (∆H°298 = –206.7 kJ mol–1).
180
Three novel bis(thiocarbonyl)disulfides with R = OSF5, Gu1 and Gu2 were explored,
and their S—S BDEs and activation barriers for homolytic S—S dissociation
calculated at the MP2/6-311+G(d,p)//B3LYP/6-31G(d) level. (F5SOCS2)2 and
((Gu2)CS2)2 were found to have S—S BDEs higher than that for TMTD, although the
reaction energy for the ((Gu2)CS2)2 case was just more endothermic than TMTD by
2.2 kJ mol–1. Similarly, activation barriers for homolytic S—S dissociation for
(F5SOCS2)2 and ((Gu2)CS2)2 were higher than TMTD, by 57.3 and 2.1 kJ mol–1,
respectively. The bis(thiocarbonyl)disulfide ((Gu1)CS2)2 showed the most promising
results, with S—S BDE at 115.9 kJ mol–1 and the corresponding barrier at 193.0 kJ
mol–1, clearly lower than that for TMTD. These calculations provide strong
theoretical evidence of ((Gu1)CS2)2 being an effective reagent in sulfur vulcanization
of rubber, which may reduce the reliance of ZnO in reactant mixtures.
References
1. Hosler, D.; Burkett, S. L.; Tarkanian, M. J. Science (Washington, DC, U. S.)
1999, 284, 1988.
2. Akiba, M.; Hashim, A. S. Prog. Polym. Sci. 1997, 22, 475.
3. Steudel, R.; Steudel, Y.; Mak, A. M.; Wong, M. W. J. Org. Chem. 2006, 71,
9302.
4. Kato, S.; Nishiwaki, M.; Inagaki, S.; Ohshima, S.; Ohno, Y.; Mizuta, M.;
Murai, T. Chem. Ber. 1985, 118, 1684.
5. Schroll, A. L.; Barany, G. J. Org. Chem. 1986, 51, 1866.
6. Chiefari, J.; Mayadunne, R. T. A.; Moad, C. L.; Moad, G.; Rizzardo, E.;
Postma, A.; Skidmore, M. A.; Thang, S. H. Macromolecules 2003, 36, 2273.
181
7. Thang, S. H.; Chong, Y. K.; Mayadunne, R. T. A.; Moad, G.; Rizzardo, E.
Tetrahedron Lett. 1999, 40, 2435.
8. Ghosh, P. J. Polym. Sci: Macromol. Rev. 1971, 5, 195.
9. Klabunde, K. J.; Kramer, M. P.; Senning, A.; Moltzen, E. K. J. Am. Chem. Soc.
1984, 106, 263.
10. Suoninen, E. J.; Thomas, T. D.; Anderson, S. E.; Runyan, M. T.; Ungier, L. J.
Electron Spectrosc. Relat. Phenom. 1985, 35, 259.
11. Victoriano, L. I. Coord. Chem. Rev. 2000, 196, 383.
12. Remko, M.; Rode, B. M. J. Mol. Struct. (Theochem) 2000, 505, 269.
13. Steudel, R.; Steudel, Y. Chem.--Eur. J. 2006, 12, 8589.
14. Adams, L. K.; Lyon, D. Y.; Alvarez, P. J. J. Water Res. 2006, 40, 3527.
15. Hirano, S.; Higo, S.; Tsukamoto, N.; Kobayashi, E.; Suzuki, K. T. Arch.
Toxicol. 1989, 63, 336.
16. Beckett, W. S.; Chalupa, D. F.; Pauly-Brown, A.; Speers, D. M.; Stewart, J. C.;
Frampton, M. W.; Utell, M. J.; Huang, L.-S.; Cox, C.; Zareba, W.;
Oberdörster, G. Am. J. Respir. Crit. Care Med. 2005, 171, 1129.
17. Seppelt, K. Angew. Chem. Int. Ed. 1982, 21, 877.
18. Kovačević, B.; Maksić, Z. B. Org. Lett. 2001, 3, 1523.
19. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.; Pople, J. A. J. Chem. Phys.
2001, 114, 108.
20. Curtiss, L. A.; Carpenter, J. E.; Raghavachari, K.; Pople, J. A. J. Chem. Phys.
1992, 96, 9030.
21. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Montgomery, J., J. A.; Vreven, T.; Kudin, K. N.; Burant, J.
C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi,
182
M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara,
M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda,
Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.;
Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R.
E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala,
P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.;
Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.;
Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.;
Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.;
Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M.
W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A.
Gaussian 03, Gaussian, Inc.: Wallingford, CT, 2004.
22. Werner, H.-J.; Knowles, P. J. MOLPRO 2002, University of Birmingham:
Birmingham, 2002.
23. Driess, M.; Merz, K.; Monsé, C. Z. Anorg. Allg. Chem. 2000, 626, 2264.
24. Cloke, F. G. N.; Hitchcock, P. B.; Hunnable, P.; Nixon, J. F.; Nyulászi, L.;
Niecke, E.; Thelen, V. Angew. Chem. Int. Ed. 1998, 37, 1083.
25. Steudel, Y.; Wong, M. W.; Steudel, R. Chem.--Eur. J. 2005, 11, 1281.
183
Appendix 1
A1.1 Time independence Approximation
Assuming that the wave function Ψ( x�,t) can be factorized into two functions, one
being a function of x� and one being a function of time t:
)()(),(0 tAxtx��
Ψ=Ψ (A1-1)
),(ˆ0 txH�
Ψ⇒
t
tAxi
tAxH
∂Ψ∂
=
Ψ=
)()(
)]()([ˆ�
ℏ
�
(A1-2)
Again, assuming that H is a linear operator that does not depend on t, dividing the
entire equation by )()( tAx�
Ψ gives:
t
tAxixHtA
∂∂
Ψ=Ψ)(
)()(ˆ)(�
ℏ�
)()(
)(ˆ )(
tAi
x
xH t
tA
∂∂
=ΨΨ
ℏ�
�
(A1-3)
(A1-4)
184
The expression on the left hand side of equation (2-5) is entirely dependent on x�
while that on the right hand side is entirely dependent on t. This can only be possible
when both are equal to a constant. Taking this constant to be E, the two equations that
are obtained are:
)(
)(tAE
t
tAi ⋅=
∂∂ℏ
(A1-5)
and
)()(ˆ xExH��
Ψ=Ψ (A1-6)
which is the time-independent Schrödinger equation.
A1.2 Application of the Born-Oppenheimer Approximation
The wave function of a molecular system can be approximately factorized into two
functions, one being a function of nuclear coordinates X�
, and the other being a
function of electronic coordinates x�
, with parametric dependence on the nuclear
coordinates.
);()(),( xXXxX elecnuc
�����Ψ⋅Ψ=Ψ (A1-7)
185
Using the same technique employed for equations (2-2) to (2-5), the electronic
Schrödinger equation is obtained:
)(
)()ˆˆ(
);(
);()ˆˆˆ(
X
XHHE
xX
xXHHH
nuc
nuc
pot
nucnuc
kin
nuc
elec
elec
pot
elecnuc
pot
elecelec
kin
elec �
�
��
��
Ψ
Ψ+−=
Ψ
Ψ++ −−−
)(XEelec
�=
⇒ );()();()ˆˆˆ( xXXExXHHH elecelecelec
pot
elecnuc
pot
elecelec
kin
elec
�����Ψ=Ψ++ −− (A1-
8)
As a consequence of the Born-Oppenheimer approximation, the expectation values
for the nuclear Hamiltonians, kin
nucH and pot
nucnucH −ˆ , can be added as constants to the
energy of the system to yield the electronic energy )(XEelec
�.
A1.3 Construction of the approximate wave function, Φ
The simplest wave function that can be constructed is the Hartree product – a product
of normalized, orthogonal orbital wave functions, φi,
≈Ψ )(x�
)(x�
Φ
= )()()( 2211 nn xxx�
…��
ϕϕϕ ⋅⋅⋅ (A1-9)
with the following condition for orthogonality:
186
ijji d δτϕϕ =∫ (A1-10)
The symbol δij denotes the Kronecker delta function, and equals to unity if i and j are
equal and zero otherwise. The differential dτ denotes an integration over space and
spin coordinates of the electrons.
The Hartree product is not an acceptable wave function, since it does not fulfill the
antisymmetric condition, that is, it must change sign when identical electrons are
interchanged:
),,,,,,(),,,,,,( 2121 xxxxxxxx ijji
�…
�…
�…
��…
�…
�…
�Ψ−=Ψ (A1-11)
To construct an antisymmetric wave function, a simplified function is first considered:
)()()(),,( 332211321 xfxfxfxxxF = (A1-12)
To make the above function antisymetric with respect to x1 and x2,
=),,( 321 xxxF
)()()(
)()()(
331221
332211
xfxfxf
xfxfxf
−
(A1-13)
Continuing the same process to make it antisymmetric with respect to changing x1 and
x2, x2 and x3, and x1 and x3, a determinantal function is obtained.
187
=),,( 321 xxxF
)()()()()()()()()(
)()()()()()()()()(
132231233211331221
231231133221332211
xfxfxfxfxfxfxfxfxf
xfxfxfxfxfxfxfxfxf
−−−
++
=
)()()(
)()()(
)()()(
333231
232221
131211
xfxfxf
xfxfxf
xfxfxf
(A1-14)
The above strategy can be applied to )(x�
Φ , to obtain the determinantal wave function.
Since electrons possess spin, spin functions α and β are also introduced. α(i) is defined
to be 1 if the spin of electron i is +½ (or spin up) and 0 otherwise; β(i) is defined to be
1 if the spin of electron i is –½ (or spin down) and 0 otherwise. The product of a
molecular orbital function φ with its spin function α or β is defined to be a spin orbital
function, with the electrons’ position and spin as its arguments. The wave function is
normalized by multiplying it by !/1 N .
=Φ )(xe
�
)()()()()()()()()()()()(
)2()()2()()2()()2()()2()()2()(
)1()()1()()1()()1()()1()()1()(
!
1
222211
222222222121
121212121111
NxNxNxNxNxNx
xxxxxx
xxxxxx
N
NNNNNNNN
NN
NN
βϕαϕβϕαϕβϕαϕ
βϕαϕβϕαϕβϕαϕβϕαϕβϕαϕβϕαϕ
��⋯
����⋮⋮⋱⋮⋮⋮⋮
��⋯
����
��⋯
����
(A1-15)
A1.4 The Fock Matrix
The electronic Hamiltonian in equation (2-11) and (2-12) can be decomposed into a
sum of one- and two-electron operators.
188
=eH kin
elecH pot
elecnucH −+ ˆ pot
elecelecH −+ ˆ
= ∑=
∇−N
i
i
1
2
2
1 ∑∑
= =
−N
i
M
A iA
A
r
Z
1 1
∑ ∑−
= +=
+1
1 1
1N
i
N
ij ijr
= ∑=
N
i
core
iH1
)(ˆ ∑ ∑
−
= +=
+1
1 1
1N
i
N
ij ijr
(A1-16)
The one-electron operator core Hamiltonian core
iH )(ˆ deals with the kinetic energy and
the potential energy of nuclear-electron attraction of electron i. The two-electron
operator deals with interelectronic repulsion energy.
The two-electron operator in equation (2-21) can be further decomposed into two
operators, the Coulomb and exchange operators, iJ and iK respectively. They are
defined as:
2
12
)2(1
)2(*)1(ˆ τϕϕ dr
J iii ∫= (A1-17)
)1()2(
1)2(*)1()1(ˆ
2
12
ijiji dr
K ϕτϕϕϕ
= ∫
(A1-18)
The one-electron and two-electron operators can be combined to form the Fock
operator:
189
∑=
−+=N
j
ijj
core
i KJHiF1
)( )]}ˆˆ2{ˆ)(ˆ (A1-19)
A1.5 Derivation of the Roothaan-Hall equations
The orbital wave functions φi in the Hartree-Fock equations are first expanded in a set
of fixed basis functions, χ as in equation (2-11). Putting these expressions for φi into
equation (2-20):
=∑
=
Nbasis
iFC1
ˆ
µµµ χ ( )∑ ∑
= =
−+
Nbasis N
j
jj
core
ii KJHC1 1
)( )1(ˆ)1(ˆ2ˆ
µµµ χ
= ∑
=
Nbasis
ii C1µ
µµ χε (A1-20)
Where the operators iJ and iK in this case are the following:
2
121 1
)2(1
)2(*)1(ˆ τχχ dr
CCJ nm
Nbasis
m
Nbasis
n
njmjj ∫∑ ∑= =
= (A1-21)
)1()2(
1)2(*)1(ˆ
1 1
2
12
n
Nbasis
m
Nbasis
n
mnjmjj dr
CCK χτχχχ µµ ∑ ∑ ∫= =
⋅= (A1-22)
Multiplying equation (2-21) by χν and integrating over all space,
∑ ∫∑ ∫
==
=Nbasis
ii
Nbasis
i dCdFC11
ˆ
µµνµ
µµνµ τχχετχχ
(A1-23)
190
Using Fµν to represent the integral matrix on the left hand side (Fock matrix) and Sµν
the integral matrix on the right hand side (overlap matrix) of equation (2-24), the
more familiar form of the Roothaan-Hall equations is arrived at after some
rearrangement:
∑=
=−Nbasis
ii CSF1
0)(ν
νµνµν ε ;
for µ = 1,2,…,Nbasis.
(A1-24)
With normalization conditions:
∑ ∑
= =
=Nbasis Nbasis
ii CSC1 1
1*µ ν
νµνµ (A1-25)
Using the following notation for two-electron repulsion integrals,
21
12
)2()2(*1
)1()1(*)|( ττϕϕϕϕλσµν σλνµ ddr∫∫
=
(A1-26)
The elements of the Fock matrix are:
)]|(
2
1)|[(
1 1
νσµλλσµνλ σ
λσµνµν ∑ ∑= =
−+=Nbasis Nbasis
core PHF (A1-27)
Where Pλσ is the one electron density matrix,
191
∑=
=occ
i
ii CCP1
*2 σλλσ (A1-28)
The electronic energy is given by:
)(
2
1
1 1
∑ ∑= =
+=Nbasis Nbasis
core
elec HFPEµ ν
µνµνµν (A1-29)
A1.6 Configuration Interaction, Singles (CIS), Configuration Interaction,
Doubles (CID) and Configuration Interaction (CISD) methods
Starting from the expression for full configuration interaction (FCI),
…+Φ+Φ+Φ=Ψ ∑∑∑∑
< <
occ
ji
virt
qp
pq
ij
pq
ij
occ
i
virt
p
p
i
p
iCI aaa 00 (A1-30)
The wave function corresponding to inclusion of single substitutions only would be:
∑∑ Φ+Φ=Ψocc
i
virt
p
p
i
p
iCIS aa 00 (A1-31)
Similarly, the wave function corresponding to the inclusion of double substitutions
would be:
192
∑∑< <
Φ+Φ=Ψocc
ji
virt
qp
pq
ij
pq
ijCID aa 00 (A1-32)
The CISD wave function, which includes single and double substitutions, is thus:
∑∑∑∑< <
Φ+Φ+Φ=Ψocc
ji
virt
qp
pq
ij
pq
ij
occ
i
virt
p
p
i
p
iCISD aaa 00 (A1-33)
A1.7 Quadratic Configuration Interaction
In QCI, the first three substitution operators, 1T , 2T and 3T are defined as:
p
i
pi
p
i taT ˆˆ
,
1 ∑ ⋅=
pq
ij
pqij
pq
ij taT ˆ4
1ˆ
,
2 ∑ ⋅=
pqr
ijk
pqrijk
pqr
ijk taT ˆ36
1ˆ
,
3 ∑ ⋅= (A1-34)
Where …,ˆ,ˆ pq
ij
p
i tt are the elementary substitution operators and the arrays
…,, pq
ij
p
i aa are coefficients to be determined. The QCISD equations are:
QCISDETH =⟩Φ⋅Φ⟨ 020
ˆ||
QCISD
p
i
p
i EaTTTTH =⟩Φ⋅++Φ⟨ 02121 )ˆˆˆˆ(|| pi,∀
193
QCISD
pq
ij
pq
ij EaTTTH =⟩Φ+++Φ⟨ 0
2
221
21 )ˆˆˆ1(|| qpji <<∀ ,
(A1-35)
A partial triples correction based on perturbation theory is applied to QCISD energies:
=∆ )(QCISDET utusts
T
t
D
u t
S
s
D
s
aVVaEE
∑∑∑ ∑ −
+
)(
12
0
= pqr
ijk
pqr
ijk
pqr
ijk
ijk pqr
pqr
ijkuuu )2()(
36
1 1 +− −∑∑ ∆ (A1-36)
The quantity (E0 – Et) is the triples excitation energy using the Fock Hamiltonian, sa
and ua are QCISD coefficients, and the u-array elements are:
=upqr
ijk
)||()||()||(
)||()||()||(
)||()||()||(
pqijarpijaqrija
pqkiarpkiaqrkia
pqjkarpjkaqrjka
r
i
q
i
p
i
r
i
q
i
p
i
r
i
q
i
p
i
+++
+++
++
(A1-37)
=upqr
ijk ∑
+++
+++
++
e re
jk
qe
jk
pe
jk
re
ki
qe
ki
pe
ki
re
ij
qe
ij
pe
ij
eipqaeirpaeiqra
ejpqaejrpaejqra
ekpqaekrpaekqra
)||()||()||(
)||()||()||(
)||()||()||(
∑
+++
+++
++
+m
re
km
qe
km
pe
km
re
jm
qe
jm
pq
jm
re
ij
qe
ij
pq
im
ijqmaijpmaijrma
kiqmakipmakirma
jkqmajkpmajkrma
)||()||()||(
)||()||()||(
)||()||()||(
(A1-38)
Where (ij||pq) denotes:
194
=)||( pqij 21
12
)]1()2()2()1([1
)2(*)1(* ττχχχχχχ ddr
qpqpji −
∫∫
(A1-39)
A1.8 Coupled Cluster Method With Singles and Doubles Substitution (CCSD)
equations
The CCSD method works on the following ansatz:
…++++= 32 ˆ
!3
1ˆ!2
1ˆ1)ˆexp( TTTT (A1-40)
Where T is the cluster operator:
…++= 21ˆˆˆ TTT (A1-41)
The CCSD equations are:
0)ˆˆexp(|)(| 0210 =⟩Φ⋅+−Φ⟨ TTEH CCSD
0)ˆˆexp(|)(| 021 =⟩Φ⋅+−Φ⟨ TTEH CCSD
p
i pi,∀
0)ˆˆexp(|)(| 021 =⟩Φ⋅+−Φ⟨ TTEH CCSD
pq
ij qpji <<∀ ,
(A1-42)
The triples correction to CCSD to obtain CCSD(T) energies is given as:
195
utusts
T
t
D
u t
S
s
D
s
T aVVaEE
CCSDE ∑∑∑ ∑ −
+=∆
)(
1)(
0
(A1-43)
A1.9 Møller-Plesset Perturbation Energies
The initial premise of MPPT is the division of the true Hamiltonian into two parts,
VHH ˆˆˆ0 λ+= (A1-44)
Where 0H is first defined as the sum of one-electron Fock operators, V is a small
perturbation, and λ an arbitrary parameter.
The ground state wave function and the corresponding true energy of the system can
be expanded as a power series as a result of Rayleigh-Schrödinger perturbation theory.
…)3(3)2(2)1()0( Ψ+Ψ+Ψ+Ψ=Ψ λλλ (A1-45)
…)3(3)2(2)1()0( EEEEE λλλ +++= (A1-46)
Substituting these equations back into the Schrödinger equation, and equating the
coefficients of λ, a series of equations is obtained. The first three equations are shown
after some rearrangement:
0)ˆ( )0()0(
0 =Ψ− EH (A1-47)
196
)0()1()1()0(
0 )ˆ()ˆ( Ψ−=Ψ− VEEH (A1-48)
)0()2()1()1()2()0(
0 )ˆ()ˆ( Ψ+Ψ−=Ψ− EVEEH (A1-49)
Ψ(0)
is the HF wave function.
From equation (A1-47),
=⟩Ψ− )0()0(
0 |ˆ EH 0
=⟩Ψ−Ψ⟨ )0()0(
0
)0( |ˆ| EH 0
=⟩ΨΨ⟨ )0(
0
)0( |ˆ| H ⟩ΨΨ⟨ )0()0()0( |E
= E(0)
= ∑
i
iε (A1-50)
Since 0H is just the sum of one-electron Fock operators, E(0)
is simply the sum of
orbital energies.
From equation (A1-48),
=⟩Ψ− )1()0(
0 |ˆ EH ⟩Ψ− )0()1( |VE
=⟩Ψ−Ψ⟨ )1()0(
0
)0( |ˆ| EH ⟩Ψ−Ψ⟨ )0()1()0( |ˆ| VE
=⟩ΨΨ⟨−⟩ΨΨ⟨ )1()0()0()1(
0
)0( ||ˆ| EH ⟩ΨΨ⟨−⟩ΨΨ⟨ )0()0()0()0()1( |ˆ|| VE
(A1-51)
197
0H is also Hermitian, which leads to
=⟩ΨΨ⟨ )0()0()1( |E ⟩ΨΨ⟨ )0()0( |ˆ|V
=)1(E ⟩ΨΨ⟨ )0()0( |ˆ|V (A1-52)
The addition of E(0)
and E(1)
yields the HF energy, EHF.
=+ )1()0( EE ⟩ΨΨ⟨+⟩ΨΨ⟨ )0()0()0(
0
)0( |ˆ||ˆ| VH
= ⟩Ψ+Ψ⟨ )0(
0
)0( |ˆˆ| VH
= HFE (A1-53)
The first-order contribution to the wave function, Ψ(1)
, is written as a linear
combination of substituted wave functions with coefficients sa .
∑ Ψ=Ψs
ssa)1(
(A1-54)
Substituting equation (A1-54) into (A1-48),
=Ψ− ∑s
ssaEH )ˆ( )0(
0 )0()1( )ˆ( Ψ−VE
=⟩Ψ−Ψ⟨ ∑s
sst aEH |)ˆ(| )0(
0 ⟩Ψ−Ψ⟨ )0()1( |)ˆ(| VEt
=⟩Ψ−Ψ⟨∑ st
s
s EHa |)ˆ(| )0(
0 ⟩ΨΨ⟨−⟩ΨΨ⟨ )0()0()1( |ˆ|| VE tt
198
( )=⟩ΨΨ⟨−⟩ΨΨ⟨∑ stst
s
s EHa |||ˆ| )0(
0 ⟩ΨΨ⟨−⟩ΨΨ⟨ )0()0()1( |ˆ|| VE tt
(A1-55)
The left hand side of equation (A1-55) is only nonzero in the case where s=t, therefore:
=− )( )0(EEa ss ⟩ΨΨ⟨− )0(|ˆ|Vs
=∴ sa
s
s
EE
V
−
⟩ΨΨ⟨)0(
)0(|ˆ|
(A1-56)
And therefore Ψ(1)
is:
=Ψ )1( ∑ Ψ
−
⟩ΨΨ⟨
s
s
s
s
EE
V)0(
)0(|ˆ|
= ∑ −
Ψ
s s
ss
EE
V
0
0 (A1-57)
The subscript t indicates a double substitution ij→pq. Vt0 are the matrix elements
involving the perturbation operator V .
Vs0
=
∫ ∫ ΨΨ ns dddV τττ …⋯ 210ˆ
For
ij→pq,
Vs0
=
(ij||pq)
= 21
12
)]1()2()2()1([1
)2(*)1(* ττχχχχχχ ddr
qpqpji −
∫∫
(A1-58)
199
The second-order contribution to the MP contribution to the MP energy, E(2)
, can be
obtained:
E(2)
= ∑ −
−D
s s
s
EE
V
0
2
0 ||
= ∑∑∑∑< < −−+
occ
ji
virt
qp jiqp
pqij
)(
|)||(| 2
εεεε
(A1-59)
The above processes can be repeated for E(3)
, E(4)
, etc. to yield higher order MP
energies (MP3, MP4, etc.).
∑ −−
−=
D
st ts
tststs
EEEE
VVVVE
))((
)(
00
0000)3( δ
(A1-60)
=)4(E
∑ ∑
∑
−−−
−−+
−−−
D
su
SDTQ
t uts
ttutuststs
D
st ts
ttss
EEEEEE
VVVVVV
EEEE
VVVV
))()((
))((
))((
000
000000
2
00
0000
δδ
(A1-61)
A1.10 Density Functional Theory
For a system of N non relativistic interacting electrons in a nonmagnetic state, in a
field of nuclei with nuclear charges ZA and nuclear coordinates RA, the electronic
Hamiltonian can be written:
200
elecH ∑∑ ∑∑∑∇= = = >=
+−−=N
i
M
A
N
i
N
j ijiA
AN
ii rr
Z
1 1 1 11
2 1
2
1
eeNe UVT ˆˆˆ ++= (A1-62)
The expectation value of NeV – the external potential, hereby written VNe – is the
attractive potential exerted by the nuclei on the electrons.
From the Hohenberg-Kohn theorem, VNe is a unique functional of )(r�
ρ and since the
Hamiltonian H is determined by VNe, the full many particle ground state is a unique
functional of the ground state electron density )(0 r�
ρ . In other words, once )(0 r�
ρ is
known, the Hamiltonian H , the ground state wave function of the system Ψ0, and the
ground state energy E0, together with all other properties, can be determined.
Accordingly the ground state energy of the system is a functional of the ground state
electron density:
)]([ 00 rE�
ρ ∫ ++= )]([)]([)( 000 rErTrdVr eeNe
����ρρρ (A1-63)
The first term depends on the actual system, and the other two terms are independent
of N, ZA and RA. The system independent terms are grouped into a functional known
as the Hohenberg-Kohn functional, )]([ rF�
ρ .
)]([ rF�
ρ )]([)]([ rErT ee
��ρρ +=
201
⟩Ψ+Ψ⟨= |ˆˆ| eeUT (A1-64)
From Eee[ρ] the classical electron-electron coulombic repulsion term J[ρ] can be
extracted. The remainder is termed the non-classical contribution to electron-electron
interaction, Encl[ρ]. The non-classical contribution includes self-interaction correction,
exchange, and Coulomb correlation.
][ρeeE ][
)()(
2
121
12
21 ρρρ
nclErrdr
rr+= ∫∫��
��
][][ ρρ nclEJ += (A1-65)
Similarly, from T[ρ], the exact kinetic energy of a non-interacting reference system
Ts[ρ] can be extracted, and the remainder will fall into the residual part, Tc[ρ].
][][][ ρρρ cs TTT += (A1-66)
Where
⟩∇⟨−= ∑ i
N
i
isT ϕϕ ||2
1 2 (A1-67)
The expression for the energy of the interacting, real system can be written now as
][][][][][)]([ ρρρρρρ nclcsNe ETTJErE ++++=�
(A1-68)
202
Writing ENe[ρ], J[ρ] and Ts[ρ] in their explicit forms, and grouping the terms for
which there is no known explicit expression for, Tc[ρ] and Encl[ρ] into an exchange
term )]([ rEXC
�ρ , the following expression for )]([ rE�ρ is obtained.
=)]([ rE
�ρ ][][)()(
2
1)(
12
21 ρρρρ
ρ XCsNe ETr
rrrdrV +−+ ∫∫∫
����
=
)]([2
1
)(1
)(2
1
)(
2
21
2
1
12
2
1
1
2
1
1
rE
rdrdrr
r
rdrr
Z
XCii
N
i
j
N
i
N
j
i
i
N
i
M
A A
A
�
����
��
ρϕϕ
ϕϕ
ϕ
+⟩∇⟨−
+
−
∑
∑∑∫∫
∑∫∑
(A1-69)
A1.11 Common Exchange and Correlation Functionals
Listed below are some explicit forms of common exchange and correlation
functionals used in most DFT calculations.
Dirac exchange functional based on the uniform electron gas (LDA/LSDA):
rdrE Dirac
X
��3
43
1
)(4
3
2
3][ ∫
−= ρπ
ρ (A1-70)
Becke 88 (B88) exchange functional:
203
rd
xx
xrEE Dirac
X
B
X
��∫ −+
−=1
2
3
4
88
sinh61)(][
βρβρ
(A1-71)
Where
3
4
)(
|)(|)(
r
rrx
�
��
ρ
ρ∇=
Lee-Yang-Parr correlation functional:
=],[ βα ρρLYP
CE
( ) ( ) rdtt
tCCb
d
a
c
WW
WFF
�
∇+∇+++
−++
+
−
−−
−
−∫
3
1
exp18
1
9
1
222
1
22
3
83
2
3
83
2
3
5
3
1
ρββαα
αα
αα
βα
ρρρρρρ
ρρρρρ
ρ
γ
(A1-72)
Where
∇−
∇=
∇−
∇=
=
+
+−=
)()(
)(|
8
1
)()(
)(|
8
1
)3(10
3
)()(
)()(12
2
2
2
2
2
2
3
2
2
22
rr
rt
rr
rt
C
rr
rr
W
W
F
��
�
��
�
��
��
σσ
σσ
βα
βα
ρρρ
ρρρ
π
ρρ
ρργ
(A1-73)
The constants a = 0.04918, b = 0.132, c = 0.2533 and d = 0.349 are derived from the
Colle-Salvetti fit to the correlation energy of the helium atom.
Perdew-Wang 1991 GGA correlation functional
204
)),,(),(( βαβα ρρρρερ dHK += (A1-74)
Where
3/23/2
6/1
6/5
)),(1(2
1)),(1(
2
1),(
),(
3
12
1
βαζβαζβα
πρ
ρρρ
σ
βα
−++=
=
u
u
d
1
23
1
3/15
24
23
1
422
4223
1)),((
),(2exp2),(
)3(
)),((400exp)),()(
7
3)),(((),,(
))),((),(1(
))),((21ln)),((
2
1),,(
),,(),,(),,(
−
−
−
−
−=
−−−=
++
++=
+=
λρρβαιε
ιλβα
ρπ
ρρρρκβαφνβα
ιβαβαλ
βαιλρρβα
βαβαβα
βα
βαβα
βα
uA
duduZrdJ
dAdA
dAdudL
dJdLdH
001667.0
004235.0
316
09.0
3/1
−=
=
=
=
=
Z
κπ
ν
νκλι
32
2
07389.0472.0723.81
007389.0266.23568.2
1000
1)(
)()(
rrr
rrr
Zrr
+++
++⋅=
−=
θ
θφ
=),( βαε
4
1111111
2222222
4
3333333
1111111
)),())(,((),,,,,,),,((
),,,,,,),,((
))),((1))(,((),,,,,,),,((
),,,,,,),,((
βαζβαζωβα
βα
βαζβαζωβα
βα
PYXWVUTre
PYXWVUTre
c
PYXWVUTre
PYXWVUTre
−
+
−−
205
709921.1
22
2)1()1()(
),(
)(4
3),(
3/4
3/43/4
3/1
=−
−−++=
+
−=
+=
c
zzz
r
ω
βαβα
βαζ
βαπβα
+++++−=
+ )(2
11ln)1(2),,,,,,,(
12/3 pyrxrwrrvturtpyxwvutre
),,(),,(),,( βαβαβα QMQKdC +=
ssQ
N
dNdN
dNddK
ddZrdM
σπρ
ρ
λβαιε
ιλβα
ιβαβαλ
βαιλβα
ρπκβαφνβα
6/16/53/1
1
2
1
1
422
422
3/15
23/22
12
32
1),(4
exp2),(
))),((),(1(
))),((21ln25.0),,(
)(
39789467.335exp)
7
3)),(((5.0),,(
−
−
−
−
=
−
−=
++
++=
−−−=
T=[0.031091, 0.015545, 0.016887]
U=[0.21370, 0.20548, 0.11125]
V=[7.5957, 14.1189, 10.357]
W=[3.5876, 6.1977, 3.6231]
X=[1.6382, 3.3662, 0.88026]
Y=[0.49294, 0.62517, 0.49671]
P=[1,1,1]
To avoid singularities in the limit 0→S
ρ
))0,,()0,(( SS QCG ρρερ +=
206
A1.12 Perturbation Operators in the Calculation of ESR G-Tensors
The operators relevant for the calculation of ESR g-tensors are:
BSgh eZ
��⋅=
2
α
(A1-75)
BSpgh eKEZ
��⋅−=−
23
4
α
(A1-76)
∑ ∑
×−
−⋅
′=−
NUCN
A
n
j
jAj
Aj
AjNSO pRr
Rr
ZS
gh
����)(
||4 3
2α
(A1-77)
∑∑≠
−
×
−
−⋅
′−=
n
kj
n
k
j
kj
kj
jeSO prr
rrS
gh
���
���
3
2
||4
α
(A1-78)
∑∑≠
×
−
−⋅−=
n
kj
n
k
k
jk
jk
jSOO prr
rrSh
���
���
3
2
||α
(A1-79)
]}))[(()](){[(||8 3
3
BRrrSBSrRrRr
Zgh AjjjjjAj
N
A
n
j Aj
Adia
NSO
NUC ����������⋅−⋅−⋅⋅−
−
′= ∑ ∑−
α
(A1-80)
]}))[(()](){[(||
1
8 3
3
BrrrSBSrrrrr
gh jkjjjjjk
n
kj
n
k kj
dia
eSO
������������ ⋅−⋅−⋅⋅−
−
′−= ∑∑
≠−
α
(A1-81)
]}))[(()](){[(||
1
2 3
3
BrrrSBSrrrrr
h kjkjjkkj
n
kj
n
k kj
dia
SOO
������������ ⋅−⋅−⋅⋅−
−−= ∑∑
≠
α
(A1-82)
Where: hZ is the electron spin Zeeman operator; hZ−KE
is the kinetic energy correction
to the electron spin Zeeman operator; hSO−N is the (electron-nuclear) spin-orbit
operator; hSO−e is the electron-electron spin-orbit operator; hSOO is the spin-other-orbit
operator; hSO−Ndia
, hSO−edia
, hSOOdia
are the diamagnetic correction terms to hSO−N, hSO−e,
and hSOO respectively, also known as ‘gauge correction terms’. ZA is the charge of
nucleus A, the total number of nuclei is NNUC and the total number of electrons is n.
207
The terms jp�
, jS�
, jr�
correspond to the momentum, spin and position operators for
electron j.
∑=n
j
jSS��
(A1-83)
The value ge is the electronic Zeeman g-factor, ge=2.0023192778, and g′ is the
electronic spin-orbit g-factor g′= 2.0046385556.