Electron Density of Delocalized Bonds As a Universal Tool for Assessing Global and Local Effects of Chemical Resonance
Summary of Professional Accomplishments
Electron density of delocalized bonds as a universal tool for assessing global and local effects of chemical resonance
Dr. Dariusz Wojciech Szczepanik
Kraków 2021
1. Name Name and Surname: Dariusz Wojciech Szczepanik Degree: Doctor of Philosophy in Chemistry Current employment: Department of Theoretical Chemistry, Faculty of Chemistry, Jagiellonian University, Gronostajowa 2, 30-387 Kraków, Poland. ORCID: 0000-0002-2013-0617 ResearcherID: E-2787-2014 ScopusID: 36835515900 Personal website: http://www.eddb.pl/aboutme Presentation of the achievements: http://www.eddb.pl/hab
2. Diplomas, degrees conferred in specific areas of science or arts, including the name of the institution which conferred the degree, year of degree conferment, title of the PhD dissertation 2008 Master of Science Department of Computational Methods in Chemistry, Faculty of Chemistry UJ Thesis: „Entropic indices of the chemical bonds from information theory” Supervisor: dr. hab. Janusz Mrozek 2013 Doctor of Philosophy in Chemistry Department of Theoretical Chemistry, Faculty of Chemistry UJ Thesis: „Probabilistic models of the chemical bond in function spaces” (de- fended with honors) Supervisor: dr. hab. Janusz Mrozek
3. Information on employment in research institutes or faculties/departments or school of arts 2015 – 2020 Department of Theoretical Chemistry, Faculty of Chemistry, Jagiellonian Uni- versity. Position: Technician 2018 – 2020 Institute of Computational Chemistry and Catalysis, University of Girona, Posi- tion: EU-researcher (MSCA-IF, postdoc). 2020 – Department of Theoretical Chemistry, Faculty of Chemistry, Jagiellonian Uni- versity. Position: Adjunct
4. Description of the achievements, set out in art. 219 para 1 point 2 of the Act The basis for the scientific achievement entitled “Electron density of delocalized bonds as a universal tool for assessing global and local effects of chemical resonance” is a series of thematically linked publications composed of 9 scientific articles (H1-H9) published in 2014-2019 in peer-reviewed jour- nals from the JCR list, and 1 monographic chapter (H10) of 2021. The subject of scientific articles 2 selected for the habilitation cycle (H1-H9) concerns development of the theoretical basis of my orig- inal computational method that enables one to “extract” from the total electron density of a molecule (obtained from quantum-chemical calculations) the so-called electron density of delocalized bonds (EDDB), which allows visualization and quantification of different effects of electron delocalization in chemical species, regardless of their size, topology and the electronic state. Publication (H10), in turn, can be regarded as a review article. All publications included in the habilitation cycle are based on my own original research ideas, and in all of these works I am the first and the corresponding author. I am the sole author of H3, H4, and H8, while in the case of H9 and H10 the second co-author is Prof. Miquel Solà, at whose invitation these works were prepared. The high percentage estimate of my contribution to the remaining articles is guided by the criterion of the amount of time spent plan- ning research study, writing software and scripts automating calculations (partially carried out by PhD students), elaboration of the results, preparation and submission of the manuscript, improving them in the review process, etc. In addition, I am the sole author of the original model of the "migrat- ing π-cycles" introduced in H7 (and being a generalization of the well-known in the literature model of “the migrating Clar’s π-sextet”).
4.1. Motivation1 It is commonly accepted that electron delocalization in aromatic rings is linked with unusual thermo- dynamic stability by means of the π-electron bookkeeping rules, like “4n+2” and “4n” (depending on the system topology and multiplicity), known from the age-old chemistry textbooks. Although these qualitative criteria of aromaticity (and antiaromaticity) adequately relate topology, symmetry, and degeneracy of molecular orbitals (MO) in the [n]annulene-like systems predominated by covalent resonance forms at their singlet or the lowest-lying triplet states, their use in a more general context regarding topologically diversified poly- and macrocyclic species (e.g. expanded porphyrins predom- inated by ionic forms), non-Kekulé molecules (e.g. radicals), etc., is not well-founded. Over the last decades an overwhelming number of quantitative ‘measures’ of aromaticity has been proposed in the literature, based on energetic, structural, magnetic, and electronic properties of molecules, thus providing a far more accurate account of aromatic stabilization than the electron-bookkeeping crite- ria. The most commonly used quantitative criteria of aromaticity within each of these groups are: 1) the aromatic stabilization energy (ASE), which is an energetic measure of π-aromaticity that emanates from the theory of valence bonds, and it can be evaluated by means of thermodynamic data for iso- desmic, (hyper)homodesmotic or isomerization reactions (also, a multitude of schemes can be found in the literature that allows one to efficiently estimate the aromatic stabilization energies for a specific class of aromatic species); 2) the harmonic-oscillator model of aromaticity (HOMA), which is a π- aromaticity index based on structural properties of molecules (being a normalized measure of devia- tions of bond lengths in aromatic molecule from the corresponding optimum bond lengths in an ide- alized non-aromatic molecule as a reference) – HOMA is close to 0 for nonaromatic species, ap- proaches 1 for highly aromatic ones, while for potentially antiaromatic rings it usually assumes neg-
1 Formal definitions, descriptions and scientific arguments adduced in this section comes from review articles in the spe- cial issue Chem. Soc. Rev 44 (2015), original publications H5, H7, and H10, as well as the following papers: A. Stanger, Chem. Comm. 2009 (2009) 1939; R. Hoffmann, Am. Sci. 103 (2015) 18; M. Solà, Front. Chem. 5 (2017) 22. 3 ative values; 3) the nucleus-independent chemical shift (NICS), which quantifies the effective mag- netic shielding at the centroid (or above) of the aromatic ring in external magnetic field – the more negative (positive) value of NICS, the more aromatic (antiaromatic) is the molecular ring in question; 4) the multicenter index (MCI), which is a non-reference index of aromaticity that can be calculated from both the ab initio molecular wave function as well as the electron density; MCI has been shown to be superior to other aromaticity descriptors as the only one that passes a set of rigorous tests for aromaticity quantifiers designed by Prof. M. Solà (University of Girona, Spain).
Unfortunately, each of the above-mentioned aromaticity ‘measures’ has shortcomings and lim- itations that sometimes may lead to wrong predictions. The ASE seems to outwardly be the most adequate measure of global aromaticity since it can be evaluated by means of thermodynamical data and it emerges for the direct relationship between structural consequences of electron delocalization and the stability. However, designing of isodesmic and homodesmotic reaction scenarios is very dif- ficult in practice and opens the door to a lot of arbitrariness. In contrast, an unquestionable advantage of HOMA is its computational and interpretative simplicity – it allows one to straightforwardly clas- sify any molecular ring as aromatic, non-aromatic or (potentially) antiaromatic. Unfortunately, the principal problem with HOMA is the necessity of parametrization of bond lengths for an idealized reference molecule, which obviously cannot be chosen unambiguously. Consequently, the practical use of HOMA is limited to aromatic and heteroaromatic systems since the parameters for chemical bonds with metal atoms are not available. Furthermore, parametrization of HOMA should be per- formed using exactly the same quantum-chemical method as used in calculations of equilibrium ge- ometries of the molecule under study, since routinely computed HOMA with the experimentally de- termined parameters is bound to suffer from large unsystematic errors and strong sensitivity to the choice of the basis set and the exchange-correlation functional in the density functional theory (DFT) based calculations. Aromaticity descriptors based on magnetic properties of molecules, especially NICS(0), NICS(1), and its axial component, NICS(1)zz, dissected NICS, etc., are one the most pref- erable measures of local aromaticity due to their relation to experiment; diatropic (aromatic) and paratropic (antiaromatic) ring currents indirectly manifest itself in the NMR spectra. However, the magnetic-based measures of aromaticity have also come under bitter criticism due to their complexity (NICS relies on the condensation of potentially complicated patterns of induced currents to a single number), and methodological shortcomings. Finally, the main disadvantage connected with the cal- culation of MCI is its numerical instability and computational cost which prevent using MCI to ana- lyze systems containing more than 12-14 atoms; additionally, MCI suffers from the method depend- ence – in particular, the exchange-correlation functional selection at the density functional theory (DFT) level.
To summarize, most of the aromaticity descriptors suffer from serious issues such as the arbi- trariness of choice of a reference system, lack of parametrization, ring-size extensivity issue, limited applicability, computational inefficiency, as well as methodological shortcomings and interpretative mistiness. However, introducing new aromaticity measures makes sense nowadays only if their per- formance provides an advantage over the already existing descriptors or they offer similar quality but at a significantly lower computational cost.
4
The main goal behind the development of the electron density of delocalized bonds (EDDB) method was to create possibly the most versatile tool that enables analysis of the electron delocaliza- tion in (but not only) aromatic species regardless of its type, size, topology, and the electronic state, free from all the above-mentioned issues and opening up new directions in the field, especially in the context of the excited-state aromaticity studies.
4.2. Theoretical background of the EDDB method Electron density of delocalized bonds is a part of the original method of one-electron density (ED) decomposition into ‘layers’ representing different levels of electron delocalization (H3):
ED(r) = EDLA(r) + EDLB(r) + EDDB(r), (1) with density of electrons localized on atoms (EDLA) representing inner shells, lone pairs, etc; elec- tron density of localized bonds (EDLB) representing typical (2-center 2-electron) Lewis-like bonds; EDDB, which represents electron density that cannot be assigned to atoms or bonds due to its (mul- ticenter) delocalized nature. In the basis of natural atomic orbitals (NAO), or any other representation of well-localized orthonormalized atomic orbitals, the spinless global electron density of delocalized bonds function, EDDBG(r), for a single-determinant molecular wavefunction is defined as follows (H2, H6):
(2) † ΩG EDDBG(푟) = ∑ 휒휇(푟)풟휇,휈 휒휈(푟), 휇,휈 where ΩG (3) 2 ΩG 휎 휎 ΩG,휎 휎 휎 † 휎 퓓 = 2 ∑ P [∑ 퓒푎,푏휺푎,푏 (흀푎,푏) 퓒푎,푏] P . 휎=훼,훽 푎,푏
In the above equation, Pσ (σ = α,β) stands for the σ spin-resolved charge and bond-order (CBO) ma- 휎 trix, 퓒푎,푏 is the matrix of linear coefficients of the appropriately orthogonalized σ spin-resolved two- center bond-order orbitals (2cBO) of the chemical bond Xa–Xb (obtained by diagonalization of the 휎 appropriate off-diagonal blocks of the CBO matrix), 흀푎,푏 represents the diagonal matrix collecting ΩG,휎 the corresponding 2cBO eigenvalues (occupation numbers), 휺푎,푏 is a diagonal matrix of the σ-spin bond-conjugation factors, and for an n-atomic molecular system, ΩG represents the set of all n(n-1)/2 possible atomic pairs (regardless of whether the atoms are formally bonded or not). The definition of ΩG,휎 the key matrix 휺푎,푏 is based on the bond-orbital projection (BOP) criterion developed by one of the authors, which relies on sophisticated orbital projection cascades involving 2cBOs, their 3-center counterparts (3cBO), and canonical MOs (H1, H4). According to BOP, for a typical well-localized ΩG,휎 (Lewis-like) bond Xa–Xb, all diagonal elements of the 휺푎,푏 matrix are close to zero, which means that the 2cBOs associated with this bond do not form effectively linear combinations with 2cBOs of all other bonds in a molecule. On the other hand, when the Xa–Xb bond is effectively conjugated with ΩG,휎 any other adjacent bond in the system, the 휺푎,푏 matrix has at least one element on its diagonal that approaches 1 (for systems with double and higher multifaceted aromaticity, the number of non-zero 5 diagonal elements is equal to the number of delocalization ‘channels’). The trace of such defined 퓓ΩG matrix can be straightforwardly interpreted as the population of electrons delocalized through the system of all conjugated bonds in a molecule, and as such, it can be used as a ‘measure’ of global aromaticity (H1, H3). One of the most distinctive features of the BOP technique is that one can easily restrict the set bonds/atomic pairs in ΩG giving rise to a series of different variants of global and local EDDB func- tions, as depicted in Fig. 1 (H10).
Figure 1. Isosurface contours and the corresponding populations of delocalized electrons in naphthalene by different global and local EDDB(r) functions. Source: H10.
In the case of the EDDBH(r) function, ΩH contains all possible atomic pairs in the molecule excluding hydrogen atoms from consideration. Thus, EDDBH(r) is also a global function that can be especially useful in the analysis of global π-aromaticity in organic molecules since the 2cBO involving hydrogen atoms tend to conjugate with the adjacent σ-bond orbitals noticeably increasing the delocalization in the σ-subsystem, which may sometimes lead to less precise conclusions. Moreover, the difference between EDDBG(r) and EDDBH(r) can be very useful in the identification of multivalent hydrogen interactions. For instance, as shown in Fig. 1, the net effect of eight C–H bonds in naphthalene on electron delocalization is 0.8054e, which means that only about 0.1e shared between carbon and hy- drogen atoms is delocalized due to conjugation with other σ-bonds in the system; a much higher values (usually greater than 0.3e) are expected for bonds containing hydrogen and dihydrogen bonds. In contrast to EDDBG(r) and EDDBH(r), the next two EDDB functions can be regarded as local aro- maticity measures (H10). In the case of the EDDBF(r) function, ΩF contains all possible atomic pairs in the selected molecular fragment (usually cyclic unit without H atoms), which in the case of a single 6-membered ring (6MR) in naphthalene gives rise to 6 ‧ 5 / 2 = 15 atomic pairs (the same number of linear combinations of 2cBO has to be considered within the BOP procedure). In turn, six chemical bonds are considered within the definition of the congeneric local EDDBP(r) function, which ‘measures’ electron delocalization along the selected pathway of adjacent bonds. Thus, in contrast to EDDBF(r), the EDDBP(r) function does not take into account the cross-ring delocalization effects, which for 6MR are associated with the resonance involving Dewar structures; in a sense, the relation between EDDBP and EDDBF corresponds to that between the multicenter index originally proposed
6 by Giambiagi et al. (IRing) and its averaged-over-permutations variant by Bultinck et al. (MCI). The EDDBP(r) function and the corresponding electron population (denoted simply by EDDBP) can be used to visualize and quantify local aromaticity in a wide range of molecular rings regardless of their size and topology, but also to assess macrocyclic aromaticity associated with particular delocalization pathways in larger structures like expanded porphyrins. Each of the EDDB(r) functions can easily be dissected into σ-, π- and higher-symmetry components by diagonalization of the corresponding den- sity matrix defined in Eq. 2. The resulting eigenfunctions, called the natural orbitals for bond delo- calization (NOBD), can be particularly useful in the identification of non-planar multifaceted aro- matic compounds because they do not tend toward mixing of different symmetry types like the ca- nonical MOs, even for twisted Möbius-type hetero- and metallacyclic aromatics (H9, H10). Within the current implementation of the EDDB method, the NOBDs are ordered from the high occupied to the low occupied ones, and, since the NOBDs responsible for aromaticity usually have the occupation numbers (ON) by order of magnitude greater than the rest of orbitals, the symmetry dissection of EDDB is straightforward. For instance, diagonalization of the EDDBH density matrix in naphthalene (Fig. 1) gives rise to 5 NOBDs (no 1-5) with occupation numbers from 1.5381e to 1.1907e, 29 NOBDs (no 6-34) with ONs from 0.0600e to 0.0011e, and the rest of NOBDs (no 35-228) have strictly zero ONs. By summation of the corresponding NOBD occupation numbers, one gets the fol- lowing: π-EDDBH = 6.5100e and σ-EDDBH = 0.5721e; thus, even though the σ-delocalization effects are still noticeable, naphthalene is clearly π-aromatic with about 65% (6.5e/10.0e) of electrons effec- tively delocalized in the π-subsystem (for comparison, the corresponding effectiveness of delocaliza- tion of π-electrons in benzene is close to 90%). It has to be stressed that the bond-orbital projection technique underlying the EDDB method provides a strict criterion for bonding orbitals to form linear combinations with each other, which can be directly related to the effectiveness of conjugation of double bonds resulting from different reso- nance forms contributing to the wavefunction. For instance, naphthalene can be represented by 3 different resonance forms as presented in Fig. 2 (H10). Assuming equal contribution of each reso- nance form to the wavefunction (confirmed by the high-level NBO-NRT calculations), each of 11 2 1 1 bond positions in naphthalene can be thus represented either by the set {3 C–C + 3 C=C} or {3 C–C 2 + 3 C=C}; e.g., the central bond is single in two forms (red circles) and double in one form (blue 2 1 circle), which gives {3 C–C + 3 C=C}. One should realize that the superposition of the resonance form containing a double bond in the center (blue circle) with one of the resonance forms containing a single bond in the same position (red circle) gives rise to the Clar/Randic structure with a single π- sextet, while the superposition of two resonance forms containing a single bond in the center (red circles) results in a single π-dectet circuit (Fig. 2b). In the former, the central bond can be regarded as fully delocalized (one-and-a-half bond), whereas in the latter, it does not take part in delocalization. Since there are two possible combinations of the resonance structures (two different Clar π-sextets) responsible for delocalization of the central bond and only one combination (π-dectet) responsible for its localization, the effectiveness of delocalization of this bond is about 67%; alternatively, one can say that the central bond in naphthalene has a 67% aromatic and a 33% aliphatic character (the bond is indeed noticeably longer than that in benzene, RCC = 1.43Å compared to RCC = 1.40Å, respec- tively). The same degree of delocalization features all other bond positions in the system; e.g., the four ‘rim’ bonds (connecting carbon atoms in positions α and β) can be represented by 1 double and 2 one-and-a-half (delocalized) bonds, which means they have a 67% aromatic and a 33% olefinic
7 character (RCC = 1.37Å, so the bonds are no- ticeably shorter than the one in benzene). From the energetic point of view, the second- order perturbation theory involving the natu- ral bond orbitals (NBO) predicts almost the same average stabilization effects (ca. 25 kcal/mol) associated with the delocalization of each double bond in all three resonance structures in naphthalene, which means that e.g. the delocalization of each ‘rim’ bond and the central bond contributes similarly to the aromatic stabilization energy. Therefore, it is clear that all chemical bonds in naphthalene have partially ‘aromatic’ character as they equally participate in the resonance, and the effectiveness of electron delocalization in naphthalene according to the EDDB predic- tions based on ab initio calculations (65%) is in full agreement with the classical reso- nance theory. In this context, it has to be em- phasized that the representation of aromatic- ity in naphthalene exclusively by a single π- dectet circuit or the migrating Clar’s π-sextet is incorrect as it always discriminates delo- calization effects associated with certain bond positions; thus, a reliable description of π-aromaticity in naphthalene requires both π- dectet and π-sextets circuits (Fig. 2b). More- Figure 2. a) The effectiveness of electron delocalization in over, in the presence of external magnetic naphthalene based on the classical resonance theory; b) Al- field, local diatropic ring currents associated ternative representations of the resonance forms involving with Clar’s π-sextets in naphthalene to a Clar’s π-sextet and π-dectet notation; c) Schematic represen- tation of the π-circuit cancellation in naphthalene under the large extent cancel each other (Fig. 2c), thus external magnetic field; d) 3D-plots of π-EDDBH and π- favoring the envelope π-dectet diatropic cur- ACID in naphthalene. Source: H10. rent (which is depicted in Fig. 2d by utilizing the π-component of the anisotropy of current-induced density, π-ACID). The case of naphthalene clearly shows that the correspondence between aromaticity and the ring-current diatropicity may be not strict, and, in this context, the results of the ring-current analysis in polycyclic and topologically diversified aromatics should be always interpreted with caution. To summarize, the electron density of delocalized bonds allows one to quantify and visualize the population of electrons delocalized through the system of all (global) or selected (local) conju- gated bonds in a wide range of aromatic species, and, in contrast to the induced ring-current methods, EDDB is derived from unperturbed one-electron density and, as such, it can be directly related to chemical resonance and its structural and energetic consequences.
8
4.3. Brief description of publications H1-H10 The scientometric data presented below comes from Google Scholar (accessed 20.04.2021). The im- pact factors (IF) for all articles and their total and average values are given according to the JCR 2019. The corresponding author is marked with the symbol ; the first (leading) author of the pub- lication is highlighted. The total and average IF for works H1–H9 are 22.622 and 2.514, respectively, and the total number of citations equals to 238 (181 without self-cications).
4.3.1. Original scientific article H1 D.W. Szczepanik, E.J. Zak, K. Dyduch, J. Mrozek, „Electron delocalization index based on bond order orbitals”, Chemical Physics Letters 593 (2014) 154−159. IF: 2.029. MNiSW: 70. Citations: 31, without self-citations: 22. In this work, I proposed a new way of quantifying the effect of electron delocalization in planar aromatic rings, based on the criterion of effectiveness of forming linear combinations involving the two-center orbitals of adjacent chemical bonds. This is the first article to outline the concept of bond- orbital projections between 2-center bond orbitals and their 3-center counterparts, giving rise to the straightforward interpretation of the outcomes of the proposed 1st order population analysis: the re- sulting populations of the electrons correspond to the integral over the ‘layer’ of electron density that is represented by more than a single resonance form.
Figure 3. Decomposition of the valence- shell electron density of the benzene mol- ecule into ‘layers’ representing electrons of localized bonds (~11.8e) and bonds de- localized in the ring (R) due to the chemi- cal resonance (~5.5e).
The estimated percentage of my contribution to the work is 70%: original idea, implementation (full coding of the method), preliminary calculations, analysis and elaboration of the results, preparing the first draft of the manuscript, coordination. Students EJZ and KD carried out part of the quantum- chemical calculations; JM served expertise and discussions of the substance of the proposed formal- ism as well as the selection of the test sets of molecules, and contributed to the editorial work on the final version of the manuscript.
4.3.2. Original scientific article H2 D.W. Szczepanik, M. Andrzejak, K. Dyduch, E.J. Zak, M. Makowski, G. Mazur, J. Mrozek, „A uniform approach to the description of multicenter bonding”, Physical Chemistry Chemical Physics 16 (2014) 20514−20523. IF: 3.430. MNiSW: 100. Citations: 51, without self-citations: 37. In this work, the concept of the 2-center bond-orbital projection into the multicenter bond orbitals has been further developed and extended by me to enable ‘extraction’ from the total 1-electron density the density layer that represents delocalized electrons regardless of the topology of the molecular fragment and its size. The proposed method of the electron density of delocalized bonds (EDDB) was
9 used to analyze the complex effects of electron delocalization in mono- and polycyclic aromatic hy- drocarbons, selected organometallic aromatics and molecular systems with atypical aromaticity.
Figure 4. Electron density of delocalized bonds (EDDB) in molecules with differ- entiated electronic structure, topology, and type of bond conjugation, such as charged rings, homo-aromatics, Möbius aromatics, etc.
The estimated percentage of my contribution to the work is 60%: original idea, implementation (full coding of the method), preliminary calculations, analysis and elaboration of the results, preparing figures and the first draft of the manuscript, coordination. Students EJZ and KD carried out part of the quantum-chemical calculations and prepared their preliminary elaboration; MA, MM, GM and JM served their expertise (especially as regards the selection of the test sets and discussion of the results) and language corrections during the work on the final version of the manuscript.
4.3.3. Original scientific article H3 D.W. Szczepanik, „A new perspective on quantifying electron localization and delocalization in molecular systems”, Computational and Theoretical Chemistry 1080 (2016) 33−37. IF: 1.605. MNiSW: 40. Citations: 26, without self-citations: 19. In this work, the original method of the bond-orbital projections between 2- and 3-center orbitals has been modified by introducing additional projections onto the space of the occupied canonical molec- ular orbitals. The newly proposed procedure gets rid of the non-orthogonality issue arising from solv- ing the eigenproblems for 2-center bond operators independently for each atomic pair. As a result, the new formalism enables a strict dissection of the total 1-electron density into layers representing (1) electrons do not involved in bonding (e.g. free pairs, core electrons, etc.), EDLA, (2) electrons pairs shared in typical two-center Lewis-like bonds, EDLB, and (3) the electrons delocalized between different chemical bonds, EDDB. The capabilities of the new methodology are illustrated on the ex- ample of benzene, s-triazine, borazine and anthracene molecules. It is worth noting that the unique- ness of the new method is that it does not divide the molecule into molecular fragments (which always leads to ambiguity resulting from the definition of chemical bonding), but only separates the electron density layers by the level of delocalization (thus, EDLA, EDLB, and EDDB are global functions). The estimated percentage of my contribution to the work is 100%.
Figure 5. A strict decomposition of the valence-shell electron density of piridine into layers EDLA, EDLB i EDDB.
10
4.3.4. Original scientific article H4 D.W. Szczepanik, „On the three-center orbital projection formalism within the electron density of delocalized bonds method”, Computational and Theoretical Chemistry 1100 (2017), 13−17. IF: 1.605. MNiSW: 40. Citations: 12, without self-citations: 6. In this work, I investigated an alternative projection scheme for the bond-order orbitals taking into account the through-bridges and through-space interactions, leading to a significant reduction of the redundant electrons population that appear in the EDDB analyses involving charged molecules. The new technique was implemented and tested using two sets of molecules: (1) charged cyclic hydrocar- bons and (2) a set of 25 different (hetero)aromatic molecules. The estimated percentage of my con- tribution to the work is 100%.
Figure 6. The new projection scheme between 3- and 2-center bond-order orbitals within the BOP formalism.
4.3.5. Original scientific article H5 D.W. Szczepanik, M. Solà, M. Andrzejak, B. Pawełek, J. Dominikowska, M. Kukułka, K. Dyduch, T.M. Krygowski, H. Szatylowicz, „The role of the long-range exchange corrections in the description of electron delocalization in aromatic species”, Journal of Computational Chemistry 38 (2017) 1640−1654. IF: 2.976. MNiSW: 100. Citations: 37, without self-citations: 29. In this work we examined reliability of the global and local description of aromatic rings by the EDDB method for the approximated density matrices obtained within the framework of the density functional theory involving diversified exchange-correlation functionals (XC). The results of the EDDB analysis were compared with those obtained using different aromatic indexes from magnetic (NICS and its variant), geometric (HOMA), and energetic criteria (HRCP), as well as selected electron- delocalization descriptors, such as IR, MCI, KMCI, FLU, PDI, ATI. The results clearly showed that that the description of the multicenter electron delocalization in molecules at the DFT level requires the use of XC functionals with long-range exchange corrections, which has so far been a very rare practice. Although the popular B3LYP hybrid XC functional works very well for molecules as re- gards e.g. reproduction of the experimental geometries (and is therefore suitable for calculations e.g. HOMA indices), it can lead to qualitatively incorrect evaluation of local aromatics according to elec- tron criteria in polycyclic molecules.
Figure 7. In the case of polycyclics aromatics, the DFT calculations involving local XC functionals without the- long-range exchange corrections, such as B3LYP, sys- tematically overestimate local aromaticity of particular rings, which is especially important in the case of linear acenes.
11
The estimated percentage of my contribution to the work is 50%: initial idea, preliminary study and calculations of selected aromaticity indices, analysis and elaboration of the results, preparation of the first draft of the manuscript, coordination. PhD students JD, BP, MK i KD carried out a part of the quantum-chemical calculations (including selected aromaticity descriptors) and collected and format- ted materials for SI; the other co-authors served their expertise (especially as regards the selection of the test sets of molecules and discussion of the results) and participated in editorial works on the paper.
4.3.6. Original scientific article H6 D.W. Szczepanik, M. Andrzejak, J. Dominikowska, B. Pawełek, T.M. Krygowski, H. Szatylowicz, M. Solà, „The electron density of delocalized bonds (EDDB) applied for quantifying aromaticity”, Physical Chemistry Chemical Physics 19 (2017) 28970−28981. IF: 3.430. MNiSW: 100. Citations: 39, without self-citations: 31. In this work, I proposed a new index to quantify electron delocalization in the aromatic ring, based on the EDDB method, to provide an alternative to popular indices such as NICS, HOMA, PDI, FLU or MCI, and free from many disadvantages of these indices. The correlational and factorial analysis of the EDDB-based index was performed with 15 selected aromatic indexes involving different sets of test molecules. In addition, a comprehensive EDDB benchmark study was carried out to compare EDDB with the multicenter indices MCI and KMCI. The indices have been tested to describe changes in the trend of aromaticity in the course of acetylene trimerization reactions and Diels-Alder cycload- ditive reactions, and how the ring size and heteroatoms affect the description of electron delocaliza- tion by EDDB and KMCI. The results revealed a 100% agreement between aromaticity predictions by the newly proposed index and the multicenter index, which in many situations suffers from nu- merical instability and is incomparably more computationally more expensive: the calculation of the MCI for a 10-member ring can take up to several hours, while the index based on the EDDBF and EDDBP functions is calculated in a fraction of a second regardless of the ring size.
Figure 8. A detailed comparative study of EDDB with other popular chemical aromatic indices revealed that the EDDB method describes the effects of elec- tron delocalization in a more universal way and that the EDDB predictions fully match the predictions of the MCI index, the calculation of which is in- comparably more computationally demanding.
The estimated percentage of my contribution to the work is 60%: initial idea, preliminary calculations, preparation of the input files with optimized geometries for all systems under study, analysis and elaboration of the results, preparation of the fisrt draft of the manuscript, coordination. PhD students JD and BP carried out part of the calculations of aromaticity indices and collected materials for SI; the other co-authors served their expertise and linguistic advices, helped with the design of the test sets of molecular rings and with the interpretation of the results of factorial analyses; all the co-authors participated in editorial works on the manuscript and its revisions.
12
4.3.7. Original scientific article H7 D.W. Szczepanik, M. Solà, T.M. Krygowski, H. Szatylowicz, M. Andrzejak, B. Pawelek, J. Dominikowska, M. Kukulka, K. Dyduch, „Aromaticity of acenes: the model of migrating π-circuits”, Physical Chemistry Chemical Physics 20 (2018) 13430−13436. IF: 3.430. MNiSW: 100. Citations: 17, without self-citations: 14. In this work I proposed a generalization of the model of "migrating sextet” by Clar being routinely used in describing local aromaticity of polycyclic molecules, especially linear acenes. Considering of the possibility of migrating larger π-cycles (naphthacycles, anthracycles, etc.) allowed us to explain the source of the discrepancies in the descriptions of such systems (referred to as the so-called "an- thracene problem"), and to better understand the relationship between aromatic nature of the electron density of a polycyclic molecule fragment and the frequency shifts observed experimentally in non- contact atomic force microscopy.
Figure 9. In contrast to the "migrating π-sextet" model by Clar, the extended model allowing mi- gration of larger π-cycles enables a strictly quan- titative prediction of local aromaticity in acenes, in accordance with experimental data (i.e. X-ray, NMR, nc-AFM, etc.).
The estimated percentage of my contribution to the work is 50%: initial idea, development of the theoretical model, carrying out part of the calculations, elaboration of the results, preparation of the first draft of the manuscript, coordination. PhD students BP, JD, MK, and KD performed calculations of selected aromaticity indices (only a small portion of those results was included in the manuscript); the other co-authors served their expertise and linguistic advices, shared the HPC resources and par- ticipated in editorial works on the final version of the manuscript.
4.3.8. Original scientific article H8 D.W. Szczepanik, „A simple alternative for the pseudo-π method”, International Journal of Quan- tum Chemistry 118 (2018) e25696. IF: 1.747. MNiSW: 70. Citations: 5, without self-citations: 4. In this work, I proposed the original method of calculating the multicenter electron delocalization index (as well as any other 1-electron quantity for the π-aromatic molecule), NPP, which offers ab initio accuracy at calculation cost of the Hückel method. The NPP method provides an alternative to the pseudo-π proposed in the literature, which has a number of disadvantages, such as the need to perform calculations twice (once within the original basis set of atomic orbitals, and once again in the basis of hydrogen atoms), the requirement of ring planarity, limited applicability (only aromatic hydrocarbons without heteroatoms or metal atoms), the size of the ring limited to 12-14 atoms, etc. The NPP method is free from these issues and, in combination with the EDDB method, enables au- tomated analysis of the electron delocalization pathways in the molecular systems of any size and topology; the current implementation allows the calculation of delocalization on pathways up to 10 000 atoms. The estimated percentage of my contribution to the work is 100%.
13
Figure 10. The NPP formalism enables calculations of the π-component of the multicenter electron delocalization index with excellent accuracy and at an incom- parably lower computational cost; e.g. the single-threaded calculation of KMCI
in C7H7MnCl2 takes 3-15 minutes (de- pending on the basis set), and NPP re- duces this time to about 0.05 s (regard- less of basis set and the theory level).
4.3.9. Original scientific article H9 D.W. Szczepanik, M. Solà, „Electron delocalization in planar metallacycles: Hückel or Möbius aromatic?”, ChemistryOpen 8 (2019) 219−227. IF: 2.370. MNiSW: 70. Citations: 21, without self- citations: 19.
In this work, I proposed a method of fragmentation and orbital decomposition of the EDDBF(r) func- tion to investigate the complex effects of electron delocalization in aromatic metallacycles. The re- sults showed that, contrary to common knowledge and the qualitative criteria formulated by Mauksch and Tsogoyeva, the type of d-conjugation topology of the transition metal does not depend on the number of π electrons in the system, but on the number of π-conjuged atoms forming a ring, and the maximum contribution of each atom to cyclic π-localization must not be greater than ~1 electron; in other words, the cooperativity effect within the multicenter π-electron sharing in the aromatic ring of a metallacycle forces a kind of “rivalry” between Hückel or Craig-Möbius conjugations, which en- gage different nd orbitals of the transition metal. The results allowed us to divide the metallacycles by the role of metal in the delocalization into 4 groups: Hückel aromatic (mainly 6-membered rings), Möbius aromatic (mainly 8-membered rings), the hybrid Hückel-Möbius aromatics (mainly 7-mem- bered rings), and quasi-aromatic hybrid aromatics (selected mono- and polycyclic systems containing metals such as Cu, Cd or Zn).
Figure 11. The EDDB method is particularly useful when structural, magnetic and qualitative rules of aromaticity based on topology and the π-electron bookkeeping fail.
The estimated percentage of my contribution to the work is 80%: original idea, carrying out of all the calculations, analysis and elaboration of the results, preparation of the first draft of the manuscript. MS invited me to publish in a special issue of ChemistryOpen, served his expertise (in particular, he helped in identification of the quasi-aromatic delocalization motif in the metallabicycles containing the Cu(I) atom), and participated in editorial work on the final version of the manuscript.
14
4.3.10. Original scientific article H10 D.W. Szczepanik, M. Solà, „The electron density of delocalized bonds (EDDB) as a measure of local and global aromaticity”; in: I. Fernandez (ed.) „Aromaticity: Modern Computational Methods and Applications”, Elsevier, 2021, Chapter 8, pp 259−283. IF: 0.000. MNiSW: 50. Citations: 0, with- out self-citations: 0. This work summarizes the most important theoretical development underlying the method of electron density of delocalized bonds (H1-H4), together with its subsequent modifications (an open-access invited paper in Chemistry-Methods, in preparation). The results of comparative analyses (H5-H7) has been presented and the capabilities of the new method of decomposition of EDDB into molecular fragments and orbital dissections (H9) has been briefly discussed. Also, the previously unpublished results of comparative analysis of EDDB with various electron density functions proposed in the literature to visualize electron delocalization has been presented. Additionally, the work includes a concise tutorial that enables a “quick start” for the beginners interested in using the EDDB method.
Figure 12. The 24π carboporphine within representations of EDDB, GIMIC, ACID, ELF, LOL, EDR and DORI.
The estimated percentage of my contribution to the work is 90%: the manuscript prepared mostly by me. MS invited me to joint publication, served his expertise especially in the context of selection of the material for the review, and participated in editorial work on the final version of the manuscript.
4.4. The EDDB method in scientific literature Over the past two years, dozens of works have been published (including the top chemical journals such as Nat. Chem., Angewandte, JACS, ACR, etc.), in which the EDDB method has been used to analyze the resonance structure and electron delocalization of new aromatic molecules; most of these papers have been prepared and published without me as a co-author, but with official thanks for the substantive evaluation of the manuscript in the acknowledgements section. At the same time, I have published several applicative works as a co-author in such prestigious journals as Chem. Sci., and Chem. Comm. Among others, the paper P31 published in 2020 deserves special attention, in which, together with the group of Prof. M. Solà (University of Girona) and the group of Prof. J. Zhu (Uni- versity of Xiamen), we demonstrated using advanced theoretical methods (including the EDDB method) the possibility of existence of the so-called Baird aromatics in metallic clusters. A news 15 about this work appeared in the ChemistryWorld magazine, and now the paper has the status of “hot article” (in just a few months from publication the work has already been cited a dozen times). In another important work from early 2021, P32, together with the group of Prof. M. Solà (University of Girona), the group of Prof. J. Zhu (University of Xiamen) and Prof. A. Muñoz-Castro (Universidad Autónoma de Chile), we have shown through theoretical calculations involving among others the EDDB method that even fullerenes with structural defects breaking local bonding conjugation can still sustain their aromatic character to a high degree; this work has already been cited 5 times within 2 months after publication. Finally, in the work P33 from early 2021, together with prof. H. Ottoson's group (Uppsala University) and prof. H. Bronstein's group (Cambridge University) we demonstrated among others that the paradigm shift in the study of electron structure and the properties of aromatic molecules associated with the EDDB method may be crucial for understanding the nature of electron excitation in organic molecules used in the design of materials for molecular electronics; the work has already been mentioned several dozen times (not to mention the preprints available since the end of last year). Finally, it is worth noticing that works P15, P22, P24, P25, P27, P29, and P34 utilizing the EDDB method and co-authored by me has been published in collaboration with experimentalists (2 other papers are under review and a total of 4 is in preparation). Although the results of the EDDB calculations were crucial for most of these works (in particular P22, where the EDDB method allowed us to discover a new type of aromaticity called the Möbius quasi-aromatic motif), they did not signif- icantly contribute to the development of the method itself and hence they are not included in the habilitation cycle.
4.5. What is and what is not the EDDB method In conclusion, EDDB is currently the only known method that describes in a strict way (at a given level of the theory) electron delocalization resulting from the effect of superposition of the resonance structures contributing to the wavefunction of a molecule. Thus, EDDB quantifies the direct cause of aromatic stabilization, not its indirect manifestations, such as reduction of the bond length alternation, anomalous chemical shifts, etc. A strict relation to the chemical resonance and the fact that EDDB is a global function of total electron density (i.e. a well-defined physical quantity), makes the method superior to other popular and commonly-used methods based on ring currents in the external magnetic field, which, despite being associated with experimental observables (like chemical shifts), to very limited extent describe the effect of electron delocalization (or rather the current of electron density) associated mainly with frontal 2- or 4-electron virtual excitations, so that for the systems with a com- plex topology their relationship to the actual effect of the resonance/aromatic stabilization disappears. It should be stressed that EDDB is not yet another index of aromaticity, but it is a universal method for assessment and quantification of the electron delocalization effects which feature the vast majority of known organic molecules and determine their structure and physico-chemical properties. In par- ticular, the EDDB does not answer the question whether a molecule is (anti)aromatic or not; with a huge number of types of aromaticity introduced in the literature and the ways of its identification and quantification, such a question remains purely academic in nature, and the value of the possible an- swer, in the context of the real impact of aromatic (de)stabilization on the structure and properties of a molecule is often only apparent (although it certainly increases the attractiveness of publication).2
2 R. Hoffmann, „The Many Guises of Aromaticity”, Am. Sci. 103 (2015) 18. 16
5. Presentation of significant scientific or artistic activity carried out at more than one university, scientific or cultural institution, especially at foreign institutions The main field of my current scientific activity is mathematical chemistry, machine learning and the development of computational methods in quantum chemistry. During my master's and doctoral stud- ies, I was primarily involved in the first of these branches, in particular the application of information theory to investigate the electron structure of chemical species. In my dissertation entitled “Probabil- istic models of the chemical bond in function spaces" I studied the properties of the purely bayesian descriptions of the electron structure and reactivity of molecules based on the theory of homogeneous Markov’s chains and the so-called orbital communication theory. My long-term scientific advisor, mentor and supervisor of my MSc and PhD theses dr. hab. J. Mrozek gave me a carte blanche in choosing the subject of my researches and developing my scientific interests from the very beginning. Consequently, out of a total of 16 works published jointly with my supervisor (M1, P1-P14, and P18), I am the first and the sole corresponding author in14 of them (P1-P4, P6-P14, and P18); some of these works have been published in highly specialized journals with relatively low impact factor (e.g. Journal of Mathematical Chemistry), and they have been cited 307 times so far (GoogleScholar, accessed 20.04.2021). The most important achievement during the PhD period I consider the new method of calculat- ing the correlation coefficient of chemical bonds (CCCB, i.e., the quantity that provides information on the effect on specific chemical bonds in a molecule of the formation/breaking of other bonds), which is an extremely useful tool in predicting chemical reaction products and among others identi- fying weak local non-carbon interactions (P8). The use of Shannon’s algebra communication chan- nels and the so-called reduced probabilities of communication of states in four-atom homogeneous Markov’s chains to assess the “correlation” of the electron populations in a molecule, led to signifi- cant reduction of the total cost of calculation of CCCB by 3-4 orders of magnitude compared to its original definition by Prof. W.A. Goddard III (based on the covariance cascade of charge-and-bond- order operators).3 The alternative definition of the correlation coefficient of chemical bonds proposed by was utilized among others to explain the local/global (de)stabilizing effects of dihydrogen bonding on the structure of the biphenyl molecule, reconciling the contradictory findings by other researchers (P8); another article addressing this topic in collaboration with Prof.M.P. Mitoraj is in preparation. After defending my doctoral thesis, I was employed as a technician at the Department of The- oretical Chemistry (Jagiellonian University); this was partly due to a chronic disease that during this period significantly reduced my scientific activity and mobility. For the first two years after PhD, I have been developing theoretical basis and partial implementation of the EDDB method (in particular, EDDBG), which is a basis for the proposed scientific achievement. The EDDB method has been rec- ognized from the very beginning by the scientific community (work H2 received very flattering re- views, e.g. “An excellent step forward in the analysis and study of electron delocalization features. Probably the most important in the last 20 years”, “The EDDB method may become a tool of refer- ence in the field of aromaticity measurements. If so, the impact could be enormous”, etc.), and it marked the beginning of the collaboration among others with:
3 T. Yamasaki, W.A. Goddard III, “Correlation Analysis of Chemical Bonds”, J. Phys. Chem. A 102 (1998) 2919. 17
• a computational group led by Prof. M. Solà (University of Girona) – which I have visited 14 times (89 days) since 2016, and 4 times I have been a member of the group within the frame- work of different mobility programs and stipends funded by the Government of Catalonia (3 months), National Agency for Academic Exchange (12 months), and European Commission (24+2 months), and the results of this collaboration are dozens of joint publications and 1 monographic chapter; • a computational group by Prof. H. Ottosson (Uppsala University) and experimental group by Dr. H. Bronstein (Cambridge University) with whom I have been collaborating since 2018 within the framework of the research on new materials for among others the singlet fission; • an experimental group by Dr. P. Zahl (Brookhaven University) and Dr. Y. Zhang (ExxonMo- bil) with whom I have been collaborating since 2018; our research focus on deepening the understanding of the electronic resonance structure of poly- and macrocyclic aromatics based on the images from the non-contact high-resolution atomic force microscopy; • a computational group by Prof. J. Zhu (Xiamen University) with whom I have been collabo- rating since 2019 in the researches on the so-called Baird’s and the adaptive aromaticity of organometallic compounds, but also, with indigenous scientists including Prof. T.M. Krygowski (University of Warsaw, collab- oration since 2017), Prof. H. Szatylowicz (Technical University of Warsaw, collaboration since 2017), Prof. M.P. Mitoraj (Jagiellonian University, collaboration since 2017), Dr. J. Dominikowska (University of Lodz, collaboration since 2017) or GawelGroup (IChO PAN, Warsaw, collaboration since 2020). In the period from 2016 to 2021, I led a total of 4 scientifically independent and competitive re- search projects, including: NCN Sonata IX (36 months), H2020 HPC-Europa3 (3 months, mini-pro- ject), H2020 MSCA-IF (24 months), and NAWA Bekker (12 months), which were directly or indi- rectly related to the development of new methodologies and computational techniques based on the EDDB approach and their practical applications. My scientific achievements after PhD were awarded among others the START scholarship of the Foundation for Polish Science (2015), two travel schol- arships funded by the Government of Catalonia (2015-2016) and the European Commission (2016- 2017), the Maria Sklodowska-Currie individual fellowship by European Commission (2018-2020), the MNiSW scholarship for young outstanding researchers (2018-2021), and Bekker’s stipend by the National Agency for Academic Exchange (2020-2021). I have also been invited several times by research centers in Poland and abroad to give a lecture, and in 2018, as one of about 50 specialists from around the world (including Nobel laureates), I was invited to attend the IUPAC congress in Mexico organized in order to update the definition of the chemical aromaticity and antiaromaticity concepts. I am currently applying for two research grants, "Electrons move in mysterious (path)ways: from monocycles to nanorings" (NCN Opus) and "The information-entropy origins of molecular ar- omaticity" (NCN Sonata Bis), in which I planned to collaborate closely with the experimental groups by Dr. P. Zahl (Brookhaven University, USA) and Dr. Y. Zhang (ExxonMobil, USA), as well as the experimental group by Dr. H. Bronstein (University of Cambridge, UK); also, I have been declared as an external collaborator in the project applications by Prof. M. Solà at the University of Girona, Spain, and Dr. P. Zahl at the University of Brookhaven, USA. In accordance with the latter, I am also preparing the application for the individual mobility fellowship within the Polish-American Fulbright Commission programme. In the coming years, I will also consider to apply for the ERC Consolidator grant to fund research on the neural-network models that could be used to facilitate the automate 18 design of the materials for molecular electronics, based on the so-called extreme resonance entropy principle for the electron excitations in organic molecules (which I have originally proposed in my NCN Sonata Bis project proposal):
The most distinguished feature of aromatic compounds is their ability to counterbalance ‘destruc- tive’ effects of the electron excitations on the ground-state system of π-conjugated bonds by re- distribution of resonance forms in such a way that maximizes the resonance-entropy production or minimizes the resonance-entropy loss to preserve as much information contained in their π- systems as possible. ❑
Figure 13. The extreme resonance entropy principle could be used among others in the interpretation of the STM and nc-HRAFM images, in controlling the course of (hyper)homodesmotic reactions, in predicting the electronic structure of the low-lying excited states, or in the design of neural networks supporting modelling of the materials with the tuneable transport and optical properties. In the proposed project (NCN Sonata Bis), implementation of the extreme resonance entropy principle will be based on the extended bond-orbital projection technique (eBOP) analogous to the one used in the formulation of the method of electron density of delocalized bonds (EDDB).
19
List of scientific or artistic achievements which present a major con- tribution to the development of a specific discipline
I. INFORMATION ON SCIENTIFIC OR ARTISTIC ACHIEVEMENTS SET OUT IN ART. 219 PARA 1. POINT 2 OF THE ACT
2. Cycle of scientific articles related thematically, pursuant to art. 219 para 1. point 2b of the Act. After PhD: H1. D.W. Szczepanik, E.J. Zak, K. Dyduch, J. Mrozek, „Electron delocalization index based on bond order orbitals”, Chemical Physics Letters 593 (2014) 154−159. H2. D.W. Szczepanik, M. Andrzejak, K. Dyduch, E.J. Zak, M. Makowski, G. Mazur, J. Mrozek, „A uniform approach to the description of multicenter bonding”, Physical Chemistry Chemical Phys- ics 16 (2014) 20514−20523. H3. D.W. Szczepanik, „A new perspective on quantifying electron localization and delocalization in molecular systems”, Computational and Theoretical Chemistry 1080 (2016) 33−37. H4. D.W. Szczepanik, „On the three-center orbital projection formalism within the electron density of delocalized bonds method”, Computational and Theoretical Chemistry 1100 (2017), 13−17. H5. D.W. Szczepanik, M. Solà, M. Andrzejak, B. Pawełek, J. Dominikowska, M. Kukułka, K. Dyduch, T.M. Krygowski, H. Szatylowicz, „The role of the long-range exchange corrections in the description of electron delocalization in aromatic species”, Journal of Computational Chemistry 38 (2017) 1640−1654. H6. D.W. Szczepanik, M. Andrzejak, J. Dominikowska, B. Pawełek, T.M. Krygowski, H. Szatylowicz, M. Solà, „The electron density of delocalized bonds (EDDB) applied for quantifying aromaticity”, Physical Chemistry Chemical Physics 19 (2017) 28970−28981. H7. D.W. Szczepanik, M. Solà, T.M. Krygowski, H. Szatylowicz, M. Andrzejak, B. Pawelek, J. Dominikowska, M. Kukulka, K. Dyduch, „Aromaticity of acenes: the model of migrating π-circuits”, Physical Chemistry Chemical Physics 20 (2018) 13430−13436. H8. D.W. Szczepanik, „A simple alternative for the pseudo-π method”, International Journal of Quantum Chemistry 118 (2018) e25696. H9. D.W. Szczepanik, M. Solà, „Electron delocalization in planar metallacycles: Hückel or Mö- bius aromatic?”, ChemistryOpen 8 (2019) 219−227. H10. D.W. Szczepanik, M. Solà, „The electron density of delocalized bonds (EDDB) as a measure of local and global aromaticity”; w: I. Fernandez (ed.) „Aromaticity: Modern Computational Methods and Applications”, Elsevier, 2021, Chapter 8, pp. 271−295.
21
II. INFORMATION ON SCIENTIFIC OR ARTISTIC ACTIVITY
2. List of published chapters in scientific monographs. Before PhD: M1. R.F. Nalewajski, D.W. Szczepanik, J. Mrozek, „Bond differentiation and orbital decoupling in the orbital-communication theory of the chemical bond”; in: J.R. Sabin, E. Brandas (ed.) „Ad- vances in Quantum Chemistry vol. 61”, Elsevier, 2011, Chapter 1, pp. 1−48. After PhD: M2 (H10). D.W. Szczepanik, M. Solà, „The electron density of delocalized bonds (EDDB) as a measure of local and global aromaticity”; in: I. Fernandez (ed.) „Aromaticity: Modern Computational Methods and Applications”, Elsevier, 2021, Chapter 8, pp. 271−295.
4. List of articles published in scientific journals (including the articles not mentioned in section I.2). Before PhD: P1. D.W. Szczepanik, J. Mrozek, “Entropic bond descriptors from separated output-reduced com- munication channels in AO-resolution”, Journal of Mathematical Chemistry 49 (2011) 562-575. P2. D.W. Szczepanik, J. Mrozek, “Probing the interplay between multiplicity and ionicity of the chemical bond”, Journal of Theoretical and Computational Chemistry 10 (2011) 471-482. P3. D.W. Szczepanik, J. Mrozek, “Symmetrical orthogonalization within linear space of molecular orbitals”, Chemical Physics Letters 521 (2012) 157-160. P4. D.W. Szczepanik, J. Mrozek, “Electron population analysis using a reference minimal set of atomic orbitals”, Computational and Theoretical Chemistry 996 (2012) 103-109. P5. R.F. Nalewajski, D.W. Szczepanik, J. Mrozek, “Basis set dependence of molecular information channels and their entropic bond descriptors”, Journal of Mathematical Chemistry 50 (2012) 1437- 1457. P6. D.W. Szczepanik, J. Mrozek, “On several alternatives for Löwdin orthogonalization”, Compu- tational and Theoretical Chemistry 1008 (2013) 15-19. P7. D.W. Szczepanik, J. Mrozek, “Ground-state projected covalency index of the chemical bond”, Computational and Theoretical Chemistry 1023 (2013) 83-87. P8. D.W. Szczepanik, J. Mrozek, “Through-space and through-bridge interactions in the correla- tion analysis of chemical bonds”, Computational and Theoretical Chemistry 1026 (2013) 72-77. P9. D.W. Szczepanik, J. Mrozek, “Stationarity of electron distribution in ground-state molecular systems”, Journal of Mathematical Chemistry 51 (2013) 1388-1396. P10. D.W. Szczepanik, J. Mrozek, “On quadratic bond-order decomposition within molecular or- bital space”, Journal of Mathematical Chemistry 51 (2013) 1619-1633. P11. D.W. Szczepanik, J. Mrozek, “Minimal set of molecule-adapted atomic orbitals from maxi- mum overlap criterion”, Journal of Mathematical Chemistry 51 (2013) 2687-2698.
22
P12. D.W. Szczepanik, J. Mrozek, “Nucleophilicity index based on atomic natural orbitals”, Jour- nal of Chemistry 2013 (2013) 684134. After PhD: P13 (H1) D.W. Szczepanik, E.J. Zak, K. Dyduch, J. Mrozek, „Electron delocalization index based on bond order orbitals”, Chemical Physics Letters 593 (2014) 154−159. P14 (H2) D.W. Szczepanik, M. Andrzejak, K. Dyduch, E.J. Zak, M. Makowski, G. Mazur, J. Mrozek, „A uniform approach to the description of multicenter bonding”, Physical Chemistry Chem- ical Physics 16 (2014) 20514−20523. P15. M. Andrzejak, D.W. Szczepanik, L. Orzeł, “The lowest triplet states of bridged cis-2,2'-bithi- ophenes - theory vs experiment”, Physical Chemistry Chemical Physics 17 (2015) 5328-5337. P16 (H3). D.W. Szczepanik, „A new perspective on quantifying electron localization and delocal- ization in molecular systems”, Computational and Theoretical Chemistry 1080 (2016) 33−37. P17 (H4). D.W. Szczepanik, „On the three-center orbital projection formalism within the electron density of delocalized bonds method”, Computational and Theoretical Chemistry 1100 (2017), 13−17. P18. D.W. Szczepanik, E.J. Zak, J. Mrozek, “From quantum superposition to orbital communica- tion”, Computational and Theoretical Chemistry 1115 (2017) 80-87. P19 (H5). D.W. Szczepanik, M. Solà, M. Andrzejak, B. Pawełek, J. Dominikowska, M. Kukułka, K. Dyduch, T.M. Krygowski, H. Szatylowicz, „The role of the long-range exchange corrections in the description of electron delocalization in aromatic species”, Journal of Computational Chemistry 38 (2017) 1640−1654. P20 (H6). D.W. Szczepanik, M. Andrzejak, J. Dominikowska, B. Pawełek, T.M. Krygowski, H. Szatylowicz, M. Solà, „The electron density of delocalized bonds (EDDB) applied for quantifying aromaticity”, Physical Chemistry Chemical Physics 19 (2017) 28970−28981. P21 (H7). D.W. Szczepanik, M. Solà, T.M. Krygowski, H. Szatylowicz, M. Andrzejak, B. Pawelek, J. Dominikowska, M. Kukulka, K. Dyduch, „Aromaticity of acenes: the model of migrating π-cir- cuits”, Physical Chemistry Chemical Physics 20 (2018) 13430−13436. P22. M.P. Mitoraj, G. Mahmoudi, F.A. Afkhami, A. Castineiras, I. Garcia-Santos, A.V. Gurbanov, M. Kukulka, F. Sagan, D.W. Szczepanik, D. A. Safin, “Quasi-aromatic Möbius metal chelates”, Inorganic Chemistry 57 (2018) 4395−4408. P23 (H8). D.W. Szczepanik, „A simple alternative for the pseudo-π method”, International Journal of Quantum Chemistry 118 (2018) e25696. P24. G. Mahmoudi, F.A. Afkhami, A. Castineiras, G. Giester, I.A. Konyaeva, F.I. Zubkov, F. Qu, A. Gupta, M.P. Mitoraj, F. Sagan, D.W. Szczepanik, D.A. Safin, “Effect of solvent on the structural diversity of quasi-aromatic Möbius cadmium(II) complexes”, Crystal Growth and Design 19 (2019) 1649−1659. P25. M.P. Mitoraj, M.G. Babashkina, K. Robeyns, F. Sagan, D.W. Szczepanik, Y. Garcia, D.A. Safin, “The chameleon-like nature of anagostic interactions and metalloaromaticity in square-pla- nar nickel complexes”, Organometallics 38 (2019) 1973−1981.
23
P26 (H9). D.W. Szczepanik, M. Solà, „Electron delocalization in planar metallacycles: Hückel or Möbius aromatic?”, ChemistryOpen 8 (2019) 219−227. P27. F. A. Afkhami, G. Mahmoudi, A. Khandar, A. Gurbanov, F. Zubkov, R. Waterman, M. Babash- kina, M.P. Mitoraj, D.W. Szczepanik, D. Safin, “Structural versatility of the quasi-aromatic Mö- bius type zinc(II)-pseudohalide complexes – experimental and theoretical investigations”, RSC Ad- vances 9 (2019) 23764−23773. P28. M.P. Mitoraj, F. Sagan, D.W. Szczepanik, J.H. de Lange, A.L. Ptaszek, D.M.E. van Niekerk, I. Cukrowski, “Origin of Hydrocarbons Stability from Computational Perspective – A Case Study of Xylene Isomers”, ChemPhysChem 21 (2020) 494–502. P29. M.P. Mitoraj, D.S. Shapenova, A.A. Shiryaev, M. Bolte, M. Kukułka, D.W. Szczepanik, J. Hooper, M.G. Babashkina, G. Mahmoudi, D.A. Safin, “Resonance assisted hydrogen bonding phe- nomenon unveiled from both experiment and theory – an example of new family of ethyl n‐salicyli- deneglycinate dyes”, Chemistry - A European Journal 26 (2020) 12987−12995. P30. D. Chen, D.W. Szczepanik, J. Zhu,, M. Solà, „Probing the origin of adaptive aromaticity in 16-valence-electron metallapentalenes”, Chemistry - A European Journal 26 (2020) 12964−12971. P31. D. Chen, D.W. Szczepanik, J. Zhu,, M. Solà, „All-metal Baird aromaticity”, Chemical Com- munications 56 (2020) 12522−12525. P32. D. Chen, D.W. Szczepanik, J. Zhu, A. Muñoz-Castro, M. Solà, „Aromaticity survival in hydrofullerenes: the case of C66H4 with its π-aromatic circuits”, Chemistry - A European Journal 27 (2021) 802−808. P33. W. Zeng, O. El Bakouri, D.W. Szczepanik, H. Bronstein, H. Ottosson, ”Excited state character of CIBA-type compounds interpreted in terms of Hückel-aromaticity: a rational for singlet fission chromophore design”, Chemical Science, accepted (10.26434/chemrxiv.13515989). P34. G. Mahmoudi, M. Babashkina, W. Maniukiewicz, F.A. Afkhami, B.B. Nunna, F.I. Zubkov, A.L. Ptaszek, D.W. Szczepanik, M.P. Mitoraj, D.A. Safin,, “Solvent-induced formation of novel Ni(II) complexes derived from bis-thiosemicarbazone ligand: an insight from experimental and the- oretical investigations”, International Journal of Molecular Sciences 12 (2021) 5337. Full and up-to-date list of publications is available on: www.eddb.pl/publications
7. Information on presentations given at national or international scientific or arts conferences, including a list of lectures delivered upon invitation and plenary lectures. Before PhD: K1. (Poster) “Chemical bond indices from communication theory.” D.W. Szczepanik, Current Trends in Theoretical Chemistry V (CTTC5), Kraków, Poland, July 6-10, 2008. International conference. K2. (Poster) “Communication theory of the chemical bond.” D.W. Szczepanik, Information Tech- nologies for Chemists, Kraków, Poland, September 19-20, 2008. Local conference. K3. (Poster) “Applications of the orbital communication theory of the chemical bond.” D.W. Szczep- anik, Central European Symposium on Theoretical Chemistry VIII (CESTC8), Dobogoko, Hungary, September 25-28, 2009. International conference.
24
K4. (Poster) “IT-ionicity concepts within the orbital communication theory.” D.W. Szczepanik, HITY - Zastosowanie teorii w badaniach molekularnych, Kraków, Poland, May 18-20, 2011. Local confer. K5. (Poster) “An information-theoretic approach to the chemical bond.” D.W.Szczepanik, Central European Symposium on Theoretical Chemistry X (CESTC10), Torun, Poland, September 25-28, 2011. International conference. K6. (Poster) “Probabilistic models of the chemical bond.” D.W. Szczepanik, Current Trends in The- oretical Chemistry VI (CTTC6), Kraków, Poland, September 1-5, 2013. International conference. Additionally, 6 local seminar talks. After PhD: K7. (Oral) “The effectiveness of bond conjugation - a new criterion of aromaticity.” D.W. Szczep- anik, Central European School of Physical Organic Chemistry (CES2016), Przesieka, Poland, June 6-10, 2016. International conference. K8. (Poster) “The influence of cations on inclusion of anthracene to β-cyclodextrin.” A. Stachowicz- Kuśnierz, D.W. Szczepanik, J. Korchowiec, Current Trends in Theoretical Chemistry VII (CTTC7), Krakow, Poland, September 4-8, 2016. International conference. K9. (Invited Lecture, Poster) “Aromaticity of metallacycles: Hückel or Möbius?” D.W. Szczepanik, M. Solà, Aromaticity 2018, Riviera Maya, Mexico, November 28 - December 1, 2018. International conference. K10. (Oral) “Electron delocalization in metallacycles: Hückel or Möbius aromatic?” D.W. Szczep- anik, M. Solà, European Meeting on Physical Organic Chemistry, Spala, Poland, June 3 - 7, 2019. International conference. K11. (Poster) “From linear to circular polycyclic compounds: aromaticity study on singlet and triplet states.”, S. Escayola, D.W. Szczepanik, A. Poater, M. Solà, Tools for Chemical Bonding, Bremen, Germany, July 14 – 19, 2019. International conference. K12. (Poster) “Global and local aromaticity reversals between singlet and triplet states in expanded porphyrins.” S. Escayola, D.W. Szczepanik, A. Poater, M. Solà, First International Conference on Excited State Aromaticity and Antiaromaticity, Sigtunastiftelsen, Sweden, July 30 – August 2, 2019. International conference. K13. (Lecture) “Electron delocalization and the magnetically induced ring current in poly- and mac- rocyclic aromatics.” D.W. Szczepanik, M. Solà, First International Conference on Excited State Ar- omaticity and Antiaromaticity, Sigtunastiftelsen, Sweden, July 30 – August 2, 2019. International conference. Additionally, 10 seminar talks, including 4 invited seminar talks abroad.
8. Information on participation in organizational and scientific committees at national or international conferences, including the applicant’s function. Before PhD: O1. Executive committee member, Current Trends in Theoretical Chemistry VI (CTTC6), Kraków, Poland, September 1-5, 2013. International conference.
25
After PhD: O2. Executive committee member, Current Trends in Theoretical Chemistry VII (CTTC7), Krakow, Poland, September 4-8, 2016. International conference. O3. Executive committee member, Current Trends in Theoretical Chemistry VIII (CTTC8), Kraków, Poland, September 1-5, 2019. International conference.
9. Information on participation in the works of research teams realizing projects fi- nanced through national and international competitions, including the projects which have been completed and projects in progress, and information on the function per- formed in the team. Research projects completed before PhD: G1. (2011-2013) “Probabilistic models of the chemical bond in the function spaces and physical space”, Ministry of Science and Higher Education, WFPD UJ: K/DSC/000133+000987+001469 (PI). Research projects completed after PhD: G2. (2015-2016) „Assessing the degree of Baird and Hückel aromaticity in Hückel-Baird hybrid spe- cies by means of electronic indices”, Generalitat de Catalunya: 2014SGR931 (participant, PI: M. Solà). G3. (2016-2019) „The application of the EDDB method in the analysis of structure and reactivity of molecular systems” National Science Centre, Poland, Sonata IX: 2015/17/D/ST4/00558 (PI). G4. (2016-2017) „Theoretical description of the quasi-aromatic stabilization effects in metallacycles with different topology”, European Comission, H2020 RIA-INFRAIA-2016-1, GA: 730897, contract nr HPC17158J2 (participant – PI of a minigrant). G5. (2018-2020) „Theoretical description of the multifaceted aromaticity and resonance effects in ground- and excited-state molecular systems”, European Comission, H2020 MSCA-IF-ST-2017, GA: 797335 (PI). Ongoing research projects: G6. (2020-2021) „Theoretical study of the ground- and excited-state electronic structure of aromatic molecules with resonance-assisted hydrogen bonds”, Polish National Agency for Academic Ex- change, Bekker programme: PPN/BEK/2019/1/00219 (PI).
10. Membership in international or national organizations and scientific societies, in- cluding the functions performed by the applicant. T1. (2018-) Polish Chemical Society (Physical Organic Chemistry Section), MN: 993036, Warsaw, Poland. T2. (2018-) American Chemical Society (Physical Chemistry Division), MN: 31494884, Blacksburg, Virginia, USA. T3. (2018-) Marie Curie Alumni Association (Theoretical and Computational Chemistry Panel), Brussels, Belgium.
26
11. Information on internships completed in scientific or artistic institutions, also abroad, including the place, time and duration of the internship and its character. S1. (2015-2016) 3-month stay in the Institute of Computational Chemistry and Catalysis, in Di- MoCaT group led by Prof. M. Solà, University of Girona, Catalonia, Spain. S2. (2018-2020) 24-month stay (H2020-MSCA-IF) in the Institute of Computational Chemistry and Catalysis, in DiMoCaT group led by Prof. M. Solà, University of Girona, Catalonia, Spain. S3. (2020-2021) 12-month stay (NAWA-Bekker) in the Institute of Computational Chemistry and Catalysis, in DiMoCaT group led by Prof. M. Solà, University of Girona, Catalonia, Spain. Additionally, in the period 2015-2020 I visited University of Girona 14 times (in total 89 days).
13. Information on scientific or artistic works reviewed, in particular those published in international journals. After PhD, in the period 2016-2020, I reviewed 47 scientific papers for 22 JCR journals, including:
R1. (RSC) Physical Chemistry Chemical Physics: 13 reviews. R2. (Elsevier) Chemical Physics Letters: 6 reviews. R3. (Wiley) Chemistry - A European Journal: 3 reviews. R4. (Elsevier) Physica A: Statistical Mechanics and Its Applications: 2 reviews. R5. (Wiley) Chemistry - An Asian Journal: 2 reviews. R6. (Wiley) ChemistrySelect: 2 reviews. R7. (Wiley) International Journal of Quantum Chemistry: 2 reviews. R8. (Wiley) Journal of Physical Organic Chemistry: 2 reviews. R9. (MDPI) Molecules: 2 reviews. R10. (ACS) Organometallics: 1 review. R11. (ACS) The Journal of Organic Chemistry: 1 review. R12. (Elsevier) Computational and Theoretical Chemistry: 1 review. R13. (Elsevier) Journal of Molecular Graphics and Modelling: 1 review. R14. (Elsevier) Journal of Molecular Liquids: 1 review. R15. (Elsevier) Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy: 1 review. R16. (RSC) CrystEngComm: 1 review. R17. (RSC) RSC Advances: 1 review. R18. (Springer) Journal of Molecular Modelling: 1 review. R19. (Taylor & Francis) Molecular Physics: 1 review. R20. (Wiley) Chemistry and Biodiversity: 1 review. R21. (Wiley) Journal of Computational Chemistry: 1 review. R22. (MDPI) International Journal of Molecular Sciences: 1 review.
27
Author's personal copy
Chemical Physics Letters 593 (2014) 154–159
Contents lists available at ScienceDirect
Chemical Physics Letters
journal homepage: www.elsevier.com/locate/cplett
Electron delocalization index based on bond order orbitals ⇑ Dariusz W. Szczepanik a, , Emil Zak_ a, Karol Dyduch a, Janusz Mrozek b a Department of Theoretical Chemistry, Jagiellonian University, Ingardena 3, 30-060 Cracow, Poland b Department of Computational Methods in Chemistry, Jagiellonian University, Ingardena 3, 30-060 Cracow, Poland article info abstract
Article history: A new index of electron delocalization in atomic rings is introduced and briefly discussed. The newly pro- Received 5 December 2013 posed delocalization descriptor is defined as an atom averaged measure of the effectiveness of forming In final form 6 January 2014 linear combinations from two-center bond-order orbitals for a given sequence of bonded atomic triplets, Available online 13 January 2014 and corresponds directly to electron population analysis; it allows one to get very compact and intuitive description of p-conjugation effects without additional parametrization and calibration to the reference molecular systems. The numerical results of illustrative calculations for several typical aromatic and homoaromatic compounds seem to validate the presented methodology and definitions. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction It is the main goal of this short paper to introduce step-by- step the algorithm of calculation of a new electron delocalization Superior stability of aromatic compounds is usually considered index for cyclic and polycyclic compounds. Unlike other delocal- as a manifestation of structure resonance and cyclic delocalization ization measures, our index directly referes to condensed atomic of electrons [1–3] (historically, it was Kekule who first solved the population of electrons and therefore it provides very intuitive problem of aromatic stabilization of molecules by introducing and compact description of p-conjugation systems within the the model of superimposing resonance forms [4]). As generally framework of population analysis. The presented methodology well-known, for planar and cyclic/polycyclic molecular structures takes advantage of some previous research on bond-order orbi- with arrangement of alternating single (r) and double bonds tals done by Jug [15] as well as the recently proposed (r þ p), the effective conjugation of 2b þ 1 p-bonds through-bridge communication formalism [16,17], formulated (b ¼ 0; 1; 2; ...) gives rise to a characteristic ring of delocalized within the framework of information-theoretic [18,19] methods electrons; therefore, for symmetric species, the overall population of exploration of molecular electronic structure and reactivity of bonding p-electrons is equally distributed between constituent [20–28]. bonds of aromatic ring. There is a multitude of sophisticated measures of electron delo- calization and aromaticity itself, concerning various aspects of 2. Computational scheme molecular properties obtained from calculations as well as experi- ments. Interesting but obviously not exhaustive overviews and The complete computational procedure of evaluation of elec- benchmarks of diverse aromaticity measures can be found in [5– tron delocalization index can be described in the following five 14]). As pointed out by Katritzky et al. [6] and confirmed and gen- steps: eralized shortly afterwards by Jug and Köster [7], an aromaticity index without reference to a measurable property is essentially 1. Calculate the density matrix D within representation of orthog- irrelevant since aromaticity itself is a multidimensional phenom- onalized atom-centered basis functions jvi (e.g. polarized or ena usually emerging in magnetic, energetic and geometric proper- unpolarized sets of minimal-basis atomic orbitals (AO) [29– ties of molecules. On the other hand, the structure resonance and 36]), electron delocalization are the most commonly regarded as a pri- D ¼ CnCy; ð1Þ mary determinant of distinctive properties of aromatics [8] and can be evaluated directly from quantum-chemistry calculations where C is a matrix of LCAO coefficients for a set of natural orbitals within the framework of Molecular Orbital (MO) theory as well jui [37] and n stands for a diagonal matrix of the corresponding as the theory of Atoms-In-Molecules by Bader (e.g. [9–14]). occupation numbers. Subsequently, for a set of two adjoining chem- ical bonds, e.g. A–X and X–B, construct the following submatrices ⇑ Corresponding author. from the corresponding off-diagonal atomic blocks of elements of E-mail address: [email protected] (D.W. Szczepanik). matrix D,
0009-2614/$ - see front matter Ó 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cplett.2014.01.006 Author's personal copy
D.W. Szczepanik et al. / Chemical Physics Letters 593 (2014) 154–159 155 1=2 2 2 0DA;X 0DX;B C~ ¼ DðDyDÞ ; D ¼ xðnbÞ hfbj#bixðnbÞ ; ð11Þ DAX ¼ ; DXB ¼ ; ð2Þ 2 2 3 3 DX;A 0 DB;X 0 where the function x is defined as follows: and 0 1 1 for ni;i < s; xðni;iÞ¼ ; s a threshold value: 0DA;X 0 1 B C s ni;i for ni;i P s @ A DAXB ¼ DX;A 0DX;B : ð3Þ ð12Þ 0DB;X 0 More details and properties of this ’’physical’’ orthogonalization In above equations scheme one can find in [32–34]. y 4. Calculate the diagonal matrix b of elements defined as follows: DA;X ¼ Dl;m : l 2 A; m 2 X ¼ðDX;AÞ ; "# ! 4 y ð Þ XAX XXB Xi 1 DB;X ¼ Dl;m : l 2 B; m 2 X ¼ðDX;BÞ : bi;i ¼ Cl;iCm;i Cl;iCm;i þ 2 Cl;kCm;k ; ð13Þ l m According to the discussion given in [26], do not take into account k¼1 the direct (through-space) interaction between atoms A and B, i.e. and determining the relative phase factors of three-center bond y DA;B ¼ðDB;AÞ 0. orbitals. The preceding equation is based on the information-theo- 2. Diagonalize matrices DAX; DXB and DAXB to find the correspond- retic criterion of maximum separation of nearly degenerated eigen- ing sets of two-center bond orbitals and their occupation values of a density matrix [27]. In accordance with [27], eigenvalues numbers, b of n3 that correspond to negative values of factor (13) usually ap- y pear in a density matrix spectrum right next to their phase-reversed jf i¼jviCAX; nAX ¼ðCAXÞ DAXCAX; AX 5 countertypes and always assume nearly the same absolute values. y ð Þ jfXBi¼jviCXB; nXB ¼ðCXBÞ DXBCXB; Therefore, their contributions to the delocalized bond density effec- tively cancels out. The diagonal matrix of such signed (phase-spec- as well as the set of three-center bond orbitals and their diagonal ified) ’’populations’’ of delocalized electrons in a set of three-center occupation matrix, bond orbitals (6) is straightforwardly determined by the product y j#AXBi¼jviCAXB; nAXB ¼ðCAXBÞ DAXBCAXB: ð6Þ b del k ¼ bn3; and N3 ¼ trk: ð14Þ As pointed out by Jug [15], within the spectrum of eigenvalues of Here, Ndel stands for the effective population of electrons delocal- s s matrices of type (2), there are always bs=2c nonnegative and 3 ized in the particular triplet of bonded atoms, A–X–B. To cancel exactly the same number of nonpositive eigenvalues. The former out negative values of k with their corresponding countertypes are related to bonding (in character) two-center bond orbitals, del b b and get the positive matrix k one cas use a simple populational jfAXi and jfXBi (e.g. r2; p2), whereas the latter refer to anti-bonding a a criterion of maximum overlap; i.e, for a given negative ki;i, the posi- ones, jfAXi and jfXBi (e.g. r2; p2). Moreover, þ tive countertype ki0;i0 always satisfy the following condition: X bXs=2c bXs=2c no 2 b 2 a 2 2 X 2 jðDAXÞl;mj ¼ 2 jðnAXÞj;jj ¼ 2 jðnAXÞj;jj ¼ tr ðnAXÞ ; ~ 2 þ ~ 2 l;m k jC j þ k 0 0 jC 0 j ¼ minimum: ð15Þ j j i;i l;i i ;i l;i l X bXs0 =2c bXs0 =2c no 2 b 2 a 2 2 jðD Þ j ¼ 2 jðn Þ 0 0 j ¼ 2 jðn Þ 0 0 j ¼ tr ðn Þ ; del l;m XB l;m XB j ;j XB j ;j XB The resulting matrix k of effective contributions to population of 0 0 j j electrons delocalized in a chemical bond of type A–X–B can be used ð7Þ to calculate the corresponding contributions to population of well- and the from algebraic properties of (3) we get localized two-center bonds, X X X loc del loc loc jðD Þ j2 ¼ jðD Þ j2 þ jðD Þ j2; k ¼ nb k ; and N ¼ trk : ð16Þ l;m AXB l;m AX l;m XB l;m 3 3 l;m l;m nonono ð8Þ Obviously, the overall population Ndel from (14) is invariant due 2 2 2 3 tr ðnAXBÞ ¼ tr ðnAXÞ þ tr ðnXBÞ : to the canceling procedure described above. As a matter of fact, for our purpose this procedure can be skipped without impacting Hence, within the spectrum of eigenvalues of s s matrices of type further step, since the electron populations kloc are not required (3), there are also bs=2c nonnegative and exactly the same number in the calculation of electron delocalization index. It follows from of nonpositive eigenvalues giving rise to the corresponding bonding properties of the Jug-type bond orbitals [15] that the electron b a and anti-bonding three-center bond orbitals, # and # , b j AXBi j AXBi population of particular three-center bond orbital, ðn Þ , is al- 3 i;i respectively (r3; r3; p3; p3, etc.). ways equally distributed between atomic orbitals centered on 3. Combine both subsets of eigenvectors of matrices (2) corre- atom X and the rest of AOs, collectively. Hence, the population sponding to bonding two-center orbitals, i.e. ! of electrons localized/delocalized ’’through’’ atom X in the atomic nb 0 sequence A–X–B reads fb fb ; fb ; nb AX ; 9 j 2i ðj AXi j XBiÞ 2 b ð Þ loc loc del del 0nXB NXj3 ¼ N3 =2 and NXj3 ¼ N3 =2: ð17Þ ~b and construct the new set of orthonormal bond orbitals jf2i that 5. Calculate the electron delocalization index for the p-atomic ring resemble the most their origins (with emphasis on highly occupied Rp, given by the atomic sequence A–X1– –Xp-2–B , defined as bond orbitals) and allows one to expand the subset of three-center ! Yp 1=p orbitals as p CR ¼ nt ; ð18Þ b b ~b C~ b b t¼1 j#3i j#AXBi¼jf2i ; n3 nAXB: ð10Þ p C~ where nt represents the tth element of the following p-tuple: The LCBO matrix can be straightforwardly calculated using the following modified version of the constrained orthogonalization del del del del n ¼ N ; N ; ...; N ; N : ð19Þ procedure recently proposed and discussed in [34], AjBAX1 X1jAX1X2 Xp 2jXp 3Xp 2B BjXp 2BA Author's personal copy
156 D.W. Szczepanik et al. / Chemical Physics Letters 593 (2014) 154–159
In the preceding equation labels A,fg Xt0 and B are given explicitly cyclohexadiene molecule p-delocalization is practically not for each triplet of bonded atoms (instead of label ’’3’’) to avoid observed; every r- and p-bond has a typical structure of well local- any confusion. ized two-center orbital. Therefore, overall electron populations from Figure 1 suggest the presence of 8 chemical bonds (6 r
According to the definition (18), one can regard CR as an average and 2 p) in the R6 unit, as expected. On the other hand, benzene populational measure of the effectiveness of forming linear combi- molecule represents distinctly different situation. Here, the calcu- nations from bond-order orbitals of particular sequences of bonded lated population of p-electrons delocalized along the R6 ring comes triplets of atoms in the atomic ring R. Using the geometric mean in to about 5:5e. One should realize, that the residual population of (18) is essential to ensure the nonlocal character of such electron 0:5e is associated with relatively weak ’’across the ring’’ interac- del delocalization measure. Consequently, if there is at least one atom tions. Indeed, additional calculations of NXj3 for atomic triplets in the ring R that prevents the unrestrained flow of delocalized involving the p-coupling between carbons atoms in para positions electrons it thereby becomes the limiting one and has dramatic im- give rise to the population of nearly 0:45e, which, if combined with pact on the overall delocalization index. Thus, the definition (18) ’’along the ring’’ delocalization satisfactorily reproduces the overall seems to be superlative more adequate in quantitative description number of six pi-electrons in benzene. of delocalization in aromatic rings, especially if compared to the Figure 2 presents five the highest occupied three-center orbitals loc loc overall delocalized population in atomic ring R or even a simple and the corresponding populations NXj3 and NXj3, respectively, for arithmetic mean, selected triplet of carbon atoms in benzene and 1,3-cyclohexadi- ene. As can be easily seen, for both systems main bonding contri- Xp Xp del p del 1 p bution is due to first three orbitals (10). In the case of absence of NR ¼ nt ; and NXjR ¼ p nt ; ð20Þ t¼1 t¼1 electron delocalization in C6H8 ring, two- and three-center orbitals are merely distinguishable, whereas in C6H6 the latter form effec- respectively. In order to demonstrate the usability of the newly pro- tive combinations of two-centered functions. It should be noticed posed measures of electron delocalization in cylclic (and policyclic) that, in the case of benzene molecule, the procedure of cancelling molecules, in the next section we will present and briefly discuss re- of delocalized populations of orbitals r3 and r3 is so far effective, sults of illustrative calculations for several selected aromatic hydro- that the residual population of about 0:003e is of the order of mag- carbons (AH) and polycyclic aromatic hydrocarbons (PAH). nitude smaller than electron populations of orbitals #3 and #3. The latter should be considered as an artefacts of orthogonalization 3. Illustrative examples procedure (11) and are typical for calculations involving extended sets of basis functions. Indeed, as demonstrated elsewhere [26], the All calculations were carried out for equilibrium geometries quantitative separation of bond orbitals without any significant determined at B3LYP/cc-pVDZ/NAO theory level [38–40,29], using hybridization–orthogonalization artefacts is viable only within ab initio quantum chemistry package GAMESS [41] and NBO 6.0 representation of the effective minimal set of atomic orbitals (e.g. software [42]. [30–36]). At first, qualitative and quantitative differences in bond-order Table 1 collects the results of calculations of indices (18) and orbitals and bond densities of two representative molecular sys- (20) performed at the B3LYP/cc-pVDZ/NAO theory level for equilib- tems, benzene and 1,3-cyclohexadiene, will be shown. Figure 1 rium geometries of the following representative aromatic hydro- carbons (AH) and polycyclic aromatic hydrocarbons (PAH), presents contours of electron densities along carbon R6-rings of shown in Figure 3: cyclopropenylium cation (I), cyclobutadienyli- both molecules, qRðrÞ, obtained by appropriate combination of density-matrix layers corresponding to elements of (19) with the um dication (II), cyclobutadienide dianion (III), cyclopentadienide loc del anion (IV), benzene (V), tropylium cation (VI), homotropylium cat- canceling procedure applied; electron densities qR ðrÞ and qR ðrÞ, loc del ion (VII), cis and trans isomers of 1,3-bishomotropylium cation giving rise to localized (NR ) and delocalized (NR ) electron popula- tions, were obtained in a very similar way, but incorporating (VIIIa and VIIIb, respectively), naphthalene (IX), phenanthrene (X) also relations (14) and (16). As follows from Figure 1, in 1,3 and coronene (XI). In the PAHs case, the R-unit naming convention is as follows: every cyclic subsystem is labeled with capitals A,B,..., while the combinations of type A–B, A–B–C, ..., always set up the largest possible rings enclosing all their R-subunits; e.g., in case of
molecule (XI), the notation A–B stands for a subsystem of R10-type ring (naphthalene-like), A–B–C–G represents a subsystem of
R14-type (pyrene-like), A–B–C–D–E–F is a subsystem of R18 (coron- ene-like) and, the largest one, A–B–C–D–E denotes a subsystem of
R22-type (benzo[c,g]phenanthrene-like). Additionaly, for polycyclic del molecules IX-XI atomic delocalization indices NXj3 were calculated for selected carbon atoms and are presented in Figure 3.
3.1. The simplest AHs
According to the Hückel’s rule [43], aromatic structures I and II contain 2 delocalized p electrons, whereas compounds III-VIII are typical 6 electron p-conjugated cyclic systems. However, beyond
Figure 1. Contours and overall populations of electron bond densities qRðrÞ along expectation, values from Table 1 palpably identify cyclopropenyli- R6-rings of benzene (a) and 1,3-cyclohexadiene (b), obtained by appropriate um cation as an electron system with almost 3 delocalized combination of density-matrix layers corresponding to elements of (19), with the electrons. The explanation for this outcome refers to the loc canceling procedure applied. Electron densities giving rise to localized (qR ðrÞ) and del loc del co-occurrence of r-, (out-of-plane) piz- as well as the concomitant delocalized (qR ðrÞ) electron populations, NR and NR , respectively, were obtained in a very similar way, but by incorporating also relations (14) and (16). Method: (in-plane) pixy-delocalization effects, connected with very high B3LYP/cc-pVDZ/NAO, equilibrium geometries. accumulation of positive charge on hydrogen atoms; indeed, the Author's personal copy
D.W. Szczepanik et al. / Chemical Physics Letters 593 (2014) 154–159 157
loc del Figure 2. The first five highest occupied three-center orbitals and the corresponding populations NXj3 and NXj3, for selected triplet of carbon atoms in benzene (a) and 1,3- cyclohexadiene (b). Method: B3LYP/cc-pVDZ/NAO, equilibrium geometries.
Table 1 Electron delocalization descriptors (18) and (20) for a number of cyclic and polycyclic aromatic systems shown in Figure 3. Only one of the symmetry equivalent cyclic units denoted by capital letters are listed. More details in the text. Method: B3LYP/cc-pVDZ/NAO, equilibrium geometries.
del del Molecule Rp p del CR Molecule Rp p del CR NR NCjR NR NCjR I 1–2–3 3 2.7630 0.9210 0.9210 VIIIb 1–8–9 3 0.3433 0.1144 0.0941 II 1–2–3 3 1.5567 0.5189 0.5188 1–2–3–4–5–7–8 7 1.8981 0.2712 0.1736 1–2–3–4 4 2.1231 0.5308 0.5308 1–2–3–4–5–6–7–8 8 2.0121 0.2515 0.1603 III 1–2-3 3 1.4476 0.4825 0.4812 1–2-3–4-5–6-7–8–9 9 2.1244 0.2360 0.1506 1–2–3–4 4 2.1270 0.5317 0.5317 IX A 6 3.5294 0.5882 0.5872 IV 1–2–3 3 0.9238 0.3079 0.0838 A–B 10 5.9127 0.5913 0.5897 1–2–3–4 4 1.8067 0.4517 0.1652 X A 6 4.2813 0.7136 0.7126 1–2–3–4–5 5 4.3607 0.8721 0.8721 B 6 2.2690 0.3782 0.3675 V 1–2–3 3 0.9235 0.3078 0.0122 A–B 10 5.1837 0.5184 0.4823 1–2–3–4 4 1.9963 0.4991 0.2670 A–B–C 14 8.0986 0.5785 0.5419 1–2–3–4-5 5 2.7866 0.5573 0.1631 XI A 6 3.2798 0.5466 0.5425 1–2–3–4–5–6 6 5.5245 0.9208 0.9208 G 6 3.3866 0.5644 0.5644 VI 1–2–3 3 0.8697 0.2899 0.0236 A–B 10 5.3492 0.5349 0.5307 1–2–3–4 4 1.7528 0.4382 0.1116 A–G 10 5.5374 0.5537 0.5512 1–2–3–4–5 5 2.6147 0.5229 0.1679 A–B–G 12 6.4787 0.5399 0.5362 1–2–3–4–5–6 6 3.4793 0.5799 0.2272 A–B–C 14 7.4191 0.5299 0.5258 1–2–3–4–5–6–7 7 6.0335 0.8619 0.8619 A–G–D 14 7.6896 0.5493 0.5457 VII 1–7–8 3 0.2091 0.0697 0.0679 A–B–C–G 14 7.4191 0.5299 0.5258 1–2–3–4–5–6–7 7 4.4826 0.6404 0.5991 A–B–C–D–G 16 8.3602 0.5225 0.5181 1–2–3–4–5–6–7–8 8 4.1125 0.5141 0.3491 A–B–C–D 18 9.4885 0.5271 0.5230 VIIIa 1–8–9 3 0.3737 0.1246 0.0859 A–B–G–D–E 18 9.5707 0.5317 0.5272 1–2–3–4–5–7–8 7 1.9938 0.2848 0.1752 A–B–C–D–E–G 18 9.3008 0.5167 0.5122 1–2–3–4-5–6–7–8 8 2.1363 0.2670 0.1665 A–B–C–D–E–F 18 9.0308 0.5017 0.4976 1–2–3–4–5–6–7–8–9 9 2.2787 0.2532 0.1602 A–B–C–D–E 22 9.5204 0.4760 0.3901
natural population analysis of constituent atoms of I gives rise to confined electron delocalization along the R5 atomic ring is nearly +0.31 charge on each hydrogen atom and about +0.02 on affected by the presence of merely weak interaction between meta each carbon atom. Similar charge accumulation can be found in carbon atoms. However, the arithmetic average of delocalized elec- compound II. In this case, substantial decrease in delocalization in- tron population (20) in R5 unit recalls a typical value for small AHs dex for R4 with respect to structure I is due to strong cross-ring (cf. II, III or IX) which is mainly due to relatively strong ’’local’’ delo- delocalization of electrons between carbon atoms. This phenome- calization of electrons along the C1–C2–C3–C4 atomic chain. This del non is plainly manifested by the comparable value of index (18) observation allows one to draw the conclusion that NR as well del for R3 and R4 ring units in II and III. Slightly weaker cross-ring inter- as NXjR cannot be considered as adequate delocalization measures action is observed in III; here, doubly occupied anti-bonding LUMO since they are not capable of comparing AHs with different num- orbital effects with the overall number of about 2 delocalized elec- bers of atoms in rings. trons instead of 6. In similar manner the R5 ring delocalization in IV is slightly below the level for benzene molecule; other possible 3.2. Tropylium and its homologs rings, i.e. R3 and R4 do not exhibit any significant p-conjugation. As one could expect, in the case of benzene molecule (V), the An interesting class of resonance-stabilized organic compounds net effective delocalization spreads over entire R6 ring, whilst form homoaromatic molecules, in which p-conjugation is inter- 3 within smaller units only R4 performs noticeable value of CR, pro- rupted by one or more sp -hybridized carbon atoms (usually meth- vided predominantly by interaction of para carbons. Concurrently, ylene groups). Such cyclic hydrocarbons, regardless of their Author's personal copy
158 D.W. Szczepanik et al. / Chemical Physics Letters 593 (2014) 154–159
III III IV V
1 1 1 4 1 4 1 6 2 5 _ 2 + 2+ 2- 3 2 3 2 5 3 3 2 34 4
VI VII VIIIa VIIIb
8 9 9
1 1 1 1 7 2 7 2 8 2 8 2 + + + + 6 3 6 3 7 3 7 3
45 456 456 45
IX X XI 0.45 0.45 0.76 0.45 0.45 0.54 0.54 0.72 0.76 AB C 0.62 0.62 0.72 0.45 0.45 0.68 AB 0.76 F G C 0.62 0.62 A B 0.45 0.45 0.54 0.54 0.76 0.27 E 0.68 0.27 D 0.45 0.45 0.45 0.45
Figure 3. Skeletal formulas for several selected organic molecules: cyclopropenylium cation (I), cyclobutadienylium dication (II), cyclobutadienide dianion (III), cyclopentadienide anion (IV), benzene (V), tropylium cation (VI), homotropylium cation (VII), cis and trans isomers of 1,3-bishomotropylium cation (VIIIa and VIIIb, respectively), naphthalene (IX), phenanthrene (X) and coronene (XI). More details in the text. diversified geometry and electronic structure, reveal chemical, two methylene groups. As follows from Table 1, both isomers, cis thermodynamic, magnetic and spectroscopic properties quite sim- (VIIIa) and trans (VIIIb), differ only marginally in terms of p-delo- ilar to the typical arenes. Probably the best studied examples of calization effectiveness in the respective ring units (but, according homoaromatic compounds are homologs of tropylium cation (VI), to expectations, the cis isomer reveals slightly higher resonance- i.e. homotropylium (VII) and 1,3-bishomotropylium cations (VIIIa stabilization in R7; R8 and R9 units). As follows from comparison and VIIIb). A reference to Table 1 indicates that in the case of tro- of delocalization indices (18) for R7 in both isomers with the corre- pylium cation (VI) pz-delocalization is slightly less effective than in sponding value for R7 in (VII) we can see that the insertion of an- the benzene molecule (V). One should notice, however, that the in- other CH2 group to homotropylium cation gives rise to dramatic crease of values of the Ct–Ct+1–Ct+2 bond angles in the R7 ring in (VI) fall of CR, i.e. down to 70%; this is somewhat understandable, since, relative to the R6 in (V) give rise to somewhat amplified interaction unlike with (VII), in both isomers of 1,3-bishomotropylium cation between carbon atoms Ct and Ct+2. Consequently, unlike the case of the R7 ring is not planar. benzene molecule, in (VI) the electron delocalization index (18) monotonically increases in R3 to R7. 3.3. Three selected PAHs
Inserting a single methylene group between atoms C1 and C7 in (VI) gives rise to the homoaromatic compound (VII). In accordance The last three structures depicted in Figure 3 represent a sub- with Table 1, electron delocalization in the R3 unit of (VII) is incon- class of polycyclic aromatic compounds involving only R6-type cyc- siderably small due to disrupting influence of sp3-hybridized lic units (denoted by capital letters). A reference to Table 1 carbon atom in CH2 group on the continuous overlapping of indicates that p-delocalization in both cyclic units of naphthalene pz-orbitals. On the other hand, the value of CR for R3 in (VII) is evi- (IX), R6-type A (B) and the composite R10-type ring (A–B), is for dently larger than for the corresponding units of (V) or (VI), which about 35% less effective relative to the benzene molecule (V). p can be connected with hiperconjugation of r-bond C–H in methy- Moreover, a detailed analysis of nt from Eq. (19) (calculated for lene group with both neighbouring carbon atoms, C1 and C7. The constituent carbon atoms of each PAHs and presented in Figure comparison of delocalization indices for R7 and R8 units clearly 3) allows one to draw the conclusion that carbon atoms in posi- indicates that p-conjugation is more effective in the former one tions b are slightly more resonance-stabilized than those in posi- 3 (DCR 0:25). In other words, the sp -carbon atom in the R8 ring tions a. Somewhat different situation one can observe in the case decreases the electron delocalization relative to the R7 unit for of phenanthrene (X); here, the difference between delocalization about 42%. Concurrently, the comparison of CR for R7 in (VII) with effectiveness in atomic rings A,C and the delocalization in B is cru- the corresponding value for the largest ring in (VI) leads to the con- cial, DC 0:35 and reflects the relative reactivity of these R6-type clusion that the insertion of group CH2 into aromatic ring of type R7 units. Mainly, two carbon atoms in less resonance-stabilized atom- results in significant decrease of electron delocalization index, ic ring B distinctly easier undergo such reactions as organic oxida-
DCR 0:26, i.e. for about 30%; similar conclusions about the rel- tion to phenanthrenequinone with chromic acid, organic reduction ative aromaticities of (VI) and (VII) can be drawn on the basis of the to dihydrophenanthrene with hydrogen gas and Raney nickel, elec- bond separation energy approach [44]. trophilic halogenation to bromophenanthrene with bromine or Bishomoaromatic compounds (VIIIa) and (VIIIb) arise from sep- even ozonolysis to diphenylaldehyde. On the other hand, the most del arative interruption of conjugation of the aromatic system (VI) by aromatic carbon atoms in A and C (NXj3 ¼ 0:76) preferably undergo Author's personal copy
D.W. Szczepanik et al. / Chemical Physics Letters 593 (2014) 154–159 159
Figure 4. An illustrative example of prediction of the electrophilic attack position based on the analysis of resonance-stabilization effects in Wheland intermediates using electron delocalization index (18). The values of CA and CB are inscribed into particular rings while the values of CA B are placed above the corresponding R10-type units. Method: B3LYP/cc-pVDZ/NAO, optimized geometries. electrophilic aromatic substitution, e.g. aromatic sulfonation to congeneric indeces based on the multicenter bond-order concept phenanthrenesulfonic acids with sulfuric acid [45]. Finally, in the [9–11] as well as the Jug’s ring-current aromaticity criterion [50]. case of coronene molecule (XI) one can observe that overhelming majority of cyclic units from R6 to R18 feature quite comparable le- References vel of delocalization, similar to that in naphthalene (X). Such equal- ization of values of (18) over almost all atomic rings results directly [1] P.R. Schleyer, Chem. Rev. 101 (2001) 1115. [2] P.R. Schleyer, Chem. Rev. 105 (2005) 3433. from the high symmetry of the ‘superbenzene’ molecule (XI) and [3] A.T. Balaban, P.R. Schleyer, H.S. Rzepa, Chem. Rev. 105 (2005) 3436. majorly influences its reactivity. [4] A. Kekulé, Liebigs Ann. Chem. 162 (1872) 77. At the end, let us take a closer look at the aromatic sulfonation [5] A.R. Katritzky, K. Jug, D.C. Oniciu, Chem. Rev. 101 (2001) 1421. [6] A.R. Katritzky, P. Barczynski, G. Musumarra, D. Pisano, M. Szafran, J. Am. Chem. of naphthalene. Obviously, due to competition between thermody- Soc. 111 (1989) 7. namic and kinetic reaction control mechanisms, it is not possible to [7] K. Jug, A.M. Köster, J. Phys. Org. Chem. 4 (1991) 163. straightforwardly predict a main product the reaction only on the [8] J. Poater, M. Duran, M. Solà, B. Silvi, Chem. Rev. 105 (2005) 3911. grounds of values of indices (17) and (18) from Figure 3 and Table 1. [9] M. Giambiagi, M.S. Giambiagi, C.D.S. Silva, A.P. Figueiredoa, Phys. Chem. Chem. Phys. 2 (2000) 3381. However, as presented in Figure 4, when the reaction is carried out [10] P. Bultinck, R. Ponec, S. Van Damme, J. Phys. Org. Chem. 18 (2005) 706. at about 80 C (kinetic control mechanisms), a-naphthalenesulfonic [11] P. Bultinck, M. Rafat, R. Ponec, B. van Gheluwe, R. Carbó-Dorca, P. Popelier, J. acid is the major product. Indeed, according to the Hammond’s Phys. Chem. A 110 (2006) 7642. [12] X. Fradera, M.A. Austen, R.F.W. Bader, J. Phys. Chem. A 103 (1999) 304. Postulate [46] (which states that the relative energies of the inter- [13] J. Poater, X. Fradera, M. Duran, M. Solà, Chem. Eur. J. 9 (2003) 400. midiates should approximate the energies of the transition states [14] X. Fradera, J. Poater, S. Simon, M. Duran, M. Solà, Theor. Chem. Acc. 108 (2002) leading to the formation of the particular products), the a-arenium 214. [15] K. Jug, J. Am. Chem. 99 (1977) 7800. ion is for about 3.5 kcal/mol more stable than the corresponding b [16] R.F. Nalewajski, Int. J. Quantum Chem. 113 (2013) 766. isomer. The same conclusion we can draw from analysis of electron [17] D.W. Szczepanik, J. Mrozek, Comput. Theor. Chem. 1026 (2013) 72. delocalization indices for particular ring units in both arenium [18] C.E. Shannon, Bell. Sys. Tech. J. 27 (1948) 379. [19] N. Abramson, Information Theory and Coding, McGraw Hill Text, 1963. ions; the a isomer reveals significantly better resonance stabiliza- [20] R.F. Nalewajski, Information Theory of Molecular Systems, Elsevier, 2006. tion with respect to all three cyclic units A, B, and A–B, i.e. [21] R.F. Nalewajski, Information Origins of the Chemical Bond, Nova Science, 2010. [22] R.F. Nalewajski, D. Szczepanik, J. Mrozek, Adv. Quantum Chem. 61 (2011) 1. DCA ¼þ0:1636; DCB ¼þ0:0835 and DCA B ¼þ0:0895 (we assume [23] D. Szczepanik, J. Mrozek, J. Math. Chem. 49 (2011) 562. that aromatic substitution involve only the atomic ring B in (IX)). [24] D. Szczepanik, J. Mrozek, J. Theor. Comp. Chem. 10 (2011) 471. [25] R.F. Nalewajski, D. Szczepanik, J. Mrozek, J. Math. Chem. 50 (2012) 1437. 4. Conclusions [26] D. Szczepanik, J. Mrozek, J. Math. Chem. 51 (2013) 1619. [27] D. Szczepanik, J. Mrozek, J. Math. Chem. 51 (2013) 1388. [28] D.W. Szczepanik, J. Mrozek, Comput Theor. Chem. 1026 (2013) 72. For typical cyclic systems of p-conjugated bonds the index of [29] A.E. Reed, R.B. Weinstock, F. Weinhold, J. Chem. Phys. 83 (1985) 735. electron delocalization defined by (18) usually assumes values in [30] W.C. Lu, C.Z. Wang, M.W. Schmidt, L. Bytautas, K.M. Ho, K. Ruedenberg, J. Chem. Phys. 120 (2004) 2629. the range from 0 up to 1. One should realize, however, that for- [31] W.C. Lu, C.Z. Wang, M.W. Schmidt, L. Bytautas, K.M. Ho, K. Ruedenberg, J. mally there is no upper bound on the value of descriptor CR, i.e. Chem. Phys. 120 (2004) 2638. [32] D. Szczepanik, J. Mrozek, Chem. Phys. Lett. 521 (2012) 157. 0 6 CR < 1. As a consequence, in the case of more complicated [33] D. Szczepanik, J. Mrozek, Comput Theor. Chem. 996 (2012) 103. electronic structures involving multi-layered delocalization effects [34] D. Szczepanik, J. Mrozek, Comput. Theor. Chem. 1008 (2013) 15. of p-(e.g. in the case of co-occurence of both, of out-of-plane pz- [35] D. Szczepanik, J. Mrozek, J. Math. Chem. 51 (2013) 2687. [36] G. Knizia, J. Chem. Theory Comput. 9 (2013) 4834. conjugation and the concomitant in-plane pxy-homoconjugation [37] P.O. Löwdin, Phys. Rev. 97 (1955) 1474. in the carbobenzene molecule [48]) and even r/d-type [47], it is ex- [38] K. Kim, K.D. Jordan, J. Phys. Chem. 98 (1994) 10089. pected that CR > 1, which, obviously, should be regarded as a dis- [39] P.J. Stephens, F.J. Devlin, C.F. Chabalowski, M.J. Frisch, J. Phys. Chem. 98 (1994) tinguishing feature rather then unwanted shortcoming. Moreover, 11623. since C is a non-reference descriptor in nature, it is not equal to [40] T.H. Dunning, J. Chem. Phys. 90 (1989) 1007. R [41] M.W. Schmidt et al., J. Comput. Chem. 14 (1993) 1347. the task of describing such phenomena as anti-aromaticity and [42] E.D. Glendening, C.R. Landis, F. Weinhold, J. Comp. Chem. 34 (2013) 1429. anti-homoaromaticity [49]. Nonetheless, the results of calculations [43] E. Hückel, Z. Phys. 70 (1931) 204. for several representative aromatic compounds seem to validate [44] J.A. Berson, J.A. Jenkins, J. Am. Chem. Soc. 94 (1972) 8907. [45] Org. Synth. Coll. 2 (1943) 482. the usability of presented methodology and definitions. [46] G.S. Hammond, J. Am. Chem. Soc. 77 (1955) 334. At the moment, we are striving to implement the computa- [47] D.Y. Zubarev, B.B. Averkiev, H.J. Zhai, L.S. Wang, A.I. Boldyrev, Phys. Chem. tional procedure of evaluating the electron delocalization index Chem. Phys. 10 (2008) 257. [48] C. Lepetit, B. Silvi, R. Chauvin, J. Phys. Chem. A 107 (2003) 464. (18) and perform a comprehensive benchmarking of diverse aro- [49] H. Volz, J. Shin, J. Org. Chem. 71 (2006) 2220. maticity measures. We are especially interested in comparative [50] K. Jug, J. Am. Chem. Soc. 48 (1983) 1344. analysis of the index CR with other commonly used and somewhat PCCP
View Article Online PAPER View Journal | View Issue
A uniform approach to the description of multicenter bonding Cite this: Phys. Chem. Chem. Phys., 2014, 16, 20514 Dariusz W. Szczepanik,*a Marcin Andrzejak,a Karol Dyduch,a Emil Z˙ak,a Marcin Makowski,a Grzegorz Mazurb and Janusz Mrozekb
A novel method for investigating the multicenter bonding patterns in molecular systems by means of the so-called Electron Density of Delocalized Bonds (EDDB) is introduced and discussed. The EDDB method combines the concept of Jug’s bond-order orbitals and the indirect (‘‘through-bridge’’) inter- action formalism and opens up new opportunities for studying the interplay between different atomic Received 4th July 2014, interactions as well as their impact on both local and global resonance stabilization in systems of Accepted 10th August 2014 conjugated bonds. Using several illustrative examples we demonstrate that the EDDB approach allows DOI: 10.1039/c4cp02932a for a reliable quantitative description of diverse multicenter delocalization phenomena (with special regard to evaluation of the aromatic stabilization in molecular systems) within the framework of a consistent www.rsc.org/pccp theoretical paradigm.
1 Introduction of Bond Orders (LOBO),8,9 representing core orbitals (1c–2e), lone pairs (1c–2e) and chemical bonds (2c–2e). The overwhelming majority of chemical interactions in mole- However, many molecules cannot be adequately described cules can be described by a set of well localized two-center two- by such localized one- or two-center orbitals and the formalism electron bonds (2c–2e), i.e. standard chemical bonds. Within of multicenter bonding has to be utilized.10–15 In this context, the framework of the age-old qualitative theory of chemical intense investigations are focused on conjugated p bonds in bonding by Lewis and Langmuir, they represent pairs of aromatics, chelatoaromatics and all-metal clusters, hypervalent electrons shared by two atoms so that each attains the electron species, boranes, molecular systems with hydrogen/dihydrogen configuration of the nearest noble gas (‘‘the octet rule’’).1,2 The bonds, agostic bonds, planar tetra- and pentacoordinated carbon development of quantum-mechanical theories of the electronic atoms, etc. The concept of the Generalized Population Analysis 16–20
Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57. structure over the decades gave rise to deeper insights into bond (GPA) has successfully been used to develop the entire forming processes and provided a multitude of sophisticated panoply of the so-called Electron Sharing Indices (ESI)23–26 tools quantifying chemical bonding patterns. One of the most congeneric with the Multicenter Indices (MCI),21,22 which well-known and resoundingly successful theories is the theory of depend on the n-order Reduced Density Matrix (n-RDM).27 Molecular Orbitals (MOs),3 within which chemical bonds in The MCI approach has turned out to be especially useful in diatomic species are described by linear combinations of atom- evaluation of multicenter electron delocalization in aromatic centered functions – Atomic Orbitals (AOs). Admittedly, in the species.28–33 One should realize, however, that this multicenter general case of polyatomics the molecular orbitals do not refer to descriptor is designed to deal only with local molecular cyclic well localized 2c–2e bonds anymore being usually delocalized units of predefined size (which opens up the door to some over the whole molecule and reflecting molecular symmetry. degree of arbitrariness). As such it does not provide a compre- Fortunately, in many polyatomic molecules the chemical language hensive tool for the description of multicenter bonding in more connected with the Lewis model can be simply adopted at the extended systems. Furthermore, being the RDM-derived quan- level of modern theory by introducing the doubly-occupied Loca- tities, MCIs are related to the many-orbital joint probabilities34–40 lized Molecular Orbitals (LMOs)4–7 or so-called Localized Orbitals and so they do not correspond directly to simple electron num- bers (even though different ways of normalization of MCIs have also been introduced41). Therefore they cannot take into account the influence of other multicenter interactions, depending on a Department of Theoretical Chemistry, Jagiellonian University, Ingardena 3, the choice of the basis set (especially within the MO-approach 30-060 Cracow, Poland. E-mail: [email protected]; 42–45 Tel: +48 12 663 22 13 involving the classical Mulliken scheme ). They are also b Department of Computational Methods in Chemistry, Jagiellonian University, subject to interpretative problems e.g. if one compares the Ingardena 3, 30-060 Cracow, Poland degree of delocalization in cyclic units of different sizes
20514 | Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 This journal is © the Owner Societies 2014 View Article Online
Paper PCCP
(renormalization allows one to remedy this problem, but it can be straightforwardly distributed between all atoms in the
always demands for predetermination of ring size). Moreover, molecule, {Xa}, for large and complex molecular systems calculations of MCIs X X X can be difficult and time consuming, especially if extended N ¼ trD ¼ traD ¼ trDa;a ¼ Na: (2) a a a basis sets are used. 46 It has recently been argued that the three-center chemical In the case of one-determinant wavefunction the density matrix interactions are by far the most significant in detecting multi- is duodempotent, i.e. D =21 kDk for k 4 1, which allows one to center electron delocalization in molecular systems. A large generalize the population analysis scheme to comprise the number of studies that can be found in the literature point at whole hierarchy of multicenter electron population indices.16–20 47–50 the relatively low importance of higher-order interactions One of the most important among them is the Wiberg-type bond in quantifying multicenter chemical bonds within the language covalency index:66 of the population analysis, i.e. using the numbers of electrons. X Herein we briefly introduce the original method of multicenter XXa Xb N 1 D 2 bonding analysis (with special regard to evaluation of the ab ¼ 2 m;n ; (3) m n aromatic stabilization in molecular systems) that takes advan- 46 tage of three-center delocalization and does not suffer from directlyreferringtotheconceptofchemicalbondorder,67–71 shortcomings and limitations of the GPA-based methods. The deeply embedded in chemical intuition. new method is formulated within the framework of molecular It has originally been pointed out by Jug8 that, within the orbital theory and, just like other familiar methods, e.g. the representation of the minimal basis of atomic orbitals, the 7,51,52,54,55 Natural Bond Order (NBO) or the Adaptive Natural bond covalency index (3) can be simply decomposed into s, p 56 Density Partitioning (AdNDP) analysis, it makes use of the and higher components by solving the following eigenproblem: age-old concept of the so-called bond order orbitals, originally 0 1 proposed by Jug.8 Unlike the already existing formalisms, 0Da;b @ A y however, the proposed approach harnesses the power of indirect Dab ¼ ¼ CablabCab: (4) Dy 0 (‘‘through-bridge’’) interaction formalism57–62 that considerably a;b simplifies the analysis of multicenter bonding patterns and Indeed, the subset of eigenvectors associated with positive opens up new opportunities for the investigation of the interplay eigenvalues of D (denoted by superscript ‘‘b’’) gives rise to between different interactions and their impact on resonance ab the two-center bonding orbitals (2cBOs): stabilization. E X X Our method is to some extent inspired by the method b b b 2 zab ¼jwiCab; and lab Nab;i ¼ Nab: (5) 46 i;i originally proposed by Bridgeman and Empson. However, con- i i trary to their model which involves colored lines and triangles to % b describe delocalization, we use a visualization tool directly related In eqn (5) matrix Cab constitutes of an extension of the b to Electron Density (ED). Note that the new approach provides the rectangular matrix Cab that expands 2cBOs on the basis of all overall picture of electron delocalization and the detailed descrip- AOs. Except that Jug’s bonding orbitals are on their own very useful in probing bonding patterns of molecular systems, they
Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57. tion of delocalized electron populations in atomic resolution. It can therefore be used for quick detection of regions of increased can also be used to ‘‘reconstruct’’ the bonding part of one- b aromaticity in very large systems as well as for quantitative electron density, r (r): studies of electron delocalization, aromatic stabilization, reactivity X ð b y b b b etc. in selected molecular fragments. r ðrÞ¼ wnðrÞDm;nwmðrÞ; r ðrÞdr ¼ N : (6) m;n 2 Theoretical background Here, Nb stands for the overall number of electrons delocalized in all chemical bonds in the molecule, and the bonding density 2.1 Bonding electron density matrix Db is defined as a simple sum of density-matrix layers Firstly, let us express the one-electron density of closed-shell corresponding to all possible pairs of atoms: molecular systems, r(r), by means of basis functions {w (r)} and m X X 2 63 b 1 b b by the corresponding one-electron density matrix as follows: D ¼ 2 Cab lab Cab: (7) a X ð a b a rðrÞ¼ wy ðrÞD w ðrÞ; rðrÞdr ¼ N: (1) n m;n m Obviously, it follows directly from the normalization condition m;n b b b b b in (6) that trD = N as well as traD = Na (in the literature Na is On the basis of well atom-assigned localized orthonormal func- usually referred to as the chemical valence of atom Xa). tions, e.g. Natural Atomic Orbitals (NAOs),64,65 one can split the spinless density matrix D into diagonal and off-diagonal atomic 2.2 Electron density of delocalized bonds 72 blocks, Da,a ={Dm,n: m, n A Xa}andDa,b ={Dm,n: m A Xa, n A Xb}, It has recently been argued that an eigenproblem analo- respectively. Consequently, the overall electron population N gous to (4) can be formulated for atomic-block off-diagonal
This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 | 20515 View Article Online
PCCP Paper
density matrices representing indirect interactions of type with the following normalization condition: ð Xa–Xb–Xg: X X 0 1 d d d d r ðrÞdr ¼ traD ¼ Na ¼ N : (14) 0Da;b 0 B C a a B C B y C y Dabg ¼ B Da;b 0Db;g C ¼ CabglabgC : (8) The EDDB matrix used in above equations takes the @ A abg following form: 0Dy 0 b;g X X b 2 by Dd ¼ 2 1 Bd ; Bd ¼ C ld C ; (15) The corresponding subset of three-center bonding orbitals ab ab ab ab ab a baa b % b (3cBOs), |zabgi =|wiCabg, is crucial for determining the multi- center delocalized electron density. Without going into details where described in our previous paper,72 it should be noted that ld ={(N d )1/2d }. (16) the projection of 3cBOs onto the set of orthogonalized 2cBOs, ab ab,i i,j ~b ~b ~b |z2i (|zabi,|zbgi), followed by the procedure of canceling of d The Bab matrix describes this part of the electron density of non-bonding and mutually phase-reversed 3cBOs, allows one to the chemical bond Xa–Xb that is delocalized in a multicenter b transform labg into a diagonal matrix collecting numbers of sense with all other bonds in the molecular system. Thus, one d electrons delocalized in a 3-center sense, labg.Furthermore,for can regard the EDDB matrix as assembled from density layers any particular triatomic sequence of conjugated bonds, Xa–Xb–Xg, relating to all possible diatomic interactions in the molecule the number of electrons delocalized ‘‘through’’ atom Xb can be under consideration. calculated straightforwardly as follows: What should be noticed is that, in the case of planar molecules/ d 1 d 2 % b d 2 % b† molecular fragments, the EDDB can also be strictly dissected into N b|abg =2 tr(labg) =trb[Cabg(labg) Cabg]. (9) the ‘‘in-plane’’ and ‘‘out-of-plane’’ EDDB-layers or the density For our purposes it is of special interest to evaluate to what layers corresponding to respective symmetry components, degree electrons assigned to the chemical bond Xa–Xb partici- EDDB = EDDB + EDDB + ..., (17) pate in the overall delocalized electron population of 3cBOs s p corresponding to both subsystems of conjugated bonds, Xa–Xb–Xg by solving the corresponding eigenproblem of the density and Xg–Xa–Xb. The simplest way to get this information is to involve matrix (15). Strict separation of the symmetry components s direct projections of 3cBOs onto the subset of orthogonalized and p of the EDDB follows mainly from the facts that degen- 2cBOs corresponding to Xa–Xb as follows: eration within the spectrum of eigenvalues of the EDDB matrix DE X X 2 practically never occurs (in contrast to the ED matrix). It has to d d ~b b d 2 Nabjabg ¼ Nab;ijabg ¼ zab;ijzabg;k labg ; (10) k;k be stressed, however, that the exact separation of higher sym- i i;k metry components may not be possible always (e.g.,adissection and of s and d bonding contributions is not necessarily possible in DE an exact way53). X X 2 d d ~b b d 2 Nabjgab ¼ Nab;ijgab ¼ zab;ijzgab;k lgab ; (11) k;k i i;k 2.3 Global and local characters of EDDB-populations Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57. d d First we would like to stress that the outlined scheme of the where Nab,i|abg and Nab,i|gab stand for the populations of electrons EDDB construction is far more efficient than ‘‘multicenter originally assigned to the ith 2cBO of the bond Xa–Xb and deloca- 54–56 lized through the corresponding atomic triplets. Obviously, electron scanning’’ techniques available in other formalisms, espe- populations from preceding equations take different values cially in the case of highly accurate wavefunctions of large-sized molecular systems. In our method the multicenter bonding depending on the choice of atom Xg. Therefore, to evaluate the electron population of the ith 2cBO that is effectively delocalized density is reconstructed by means of two-atomic fragments and d regards only local resonance triatomic hybrids, Xa–Xb–Xg and in a three-center sense, N ab,i, it is necessary to calculate orbital Xg–Xa–Xb, representing the corresponding indirect interactions populations (10–11) for each possible choice of atom Xg covalently b b (‘‘Xa with Xg through Xb’’ and ‘‘Xg with Xb through Xa’’, bonded with atom Xa or Xb (i.e. g a a, b and Nag, Nbg Z t ,wheret d respectively). is an arbitrary threshold value). Then, we can define Nab,i as follows: It should be emphasized that the use of such indirect d d d a N ab,i = max{N ab,i|abg, N ab,i|gab: g a, b}. (12) interaction formalism allows one to investigate the influence of particular chemical interactions and their mutual coupling Thus, one can interpret N d as the highest number of ab,i on the effectiveness of multicenter bonding in particular mole- electrons of bonding orbital |zb i that effectively participate ab,i cular fragment by ‘‘enabling’’ or ‘‘disabling’’ the appropriate in a three-center bonding with all other atoms in the molecule. subspace of interacting atoms X for each two-atomic density Finally, making use of eqn (6)–(11) we can define the Electron g layer. Therefore, beyond a routine study of multicenter bonds Density of Delocalized Bonds (EDDB) as X and aromatic stabilization, the EDDB method provides precise d y d EDDB r ðrÞ¼ wnðrÞDm;nwmðrÞ; (13) and valuable information about the coupling between two adjacent m;n rings in polycyclic aromatics, the impact of cross-ring interactions
20516 | Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 This journal is © the Owner Societies 2014 View Article Online
Paper PCCP
It has to be emphasized that the use of the representation of natural atomic orbitals is crucial for the EDDB-based popula- tion analysis since the NAO-based populations automatically satisfy Pauli constraints. Furthermore, the stability of the weighted orthogonalization procedure used in the construction of NAOs65,81,82 automatically ensures appropriate convergence profiles and numerical stability of atomic charges and bond orders with respect to basis set enlargement.64,65,83 This contrasts sharply with electron populations and the corresponding multicenter indices obtained within the framework of Mulliken’s popula- tion analysis scheme, which are known to exhibit unphysical negative values and numerical instabilities when the extended basis sets are used.84–86
3.1 Simple aromatic hydrocarbons Fig. 2 presents isosurfaces of ED and EDDB with the corre- sponding electron populations, global (black numbers) and the Kekule´-like (bold burgundy numbers), for the following simple aromatic hydrocarbons (AH): cyclopropenyl cations d d Fig. 1 Global and local (‘‘Kekule´an’’) EDDB-populations N and NC (along- + 2+ (C3H3 ), cyclobutadienyl dications (C4H4 ), cyclopentadienyl side atoms); contributions from hydrogen atoms are neglected. Method: anions (C5H5 ), benzene (C6H6) and cycloheptatrienyl cations B3LYP/6-31G*/NAO, equilibrium geometry. + (C7H7 ); the calculations were performed at the CAM-B3LYP/
aug-cc-pVTZ/NAO theory level and at tb = 0.001 (bonding threshold)
on the effectiveness of along-ring multicenter delocalization, etc. The latter is clearly illustrated in Fig. 1 with the benzene molecule used as an example. Even a cursory look at global (including delocalized electron contributions from all possible triatomic resonance hybrids) and Kekule´-like EDDB-populations allows one to draw the conclusion that local resonances between cross-ring interactions (mainly Dewar’s para carbon–carbon bonds) and along-ring carbon–carbon bonds contribute notice- ably to the global EDDB (up to 8% of the overall delocalized electron population assigned to carbon atoms). Therefore their influence on the electron delocalization within such a 6-member
Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57. molecular cyclic unit should not be neglected, especially in accurate calculations. Indeed, it is well known that the cross- ring interactions are more important in benzenoid-like units in which the para-delocalization effect was the basis of methods such as the PDI.28,73–75
3 Several illustrative examples
To demonstrate the performance of the EDDB approach, several illustrative examples are presented and briefly discussed. All the ab initio calculations were performed using Gamess76,77 and Gaussian78 packages at the DFT level with the B3LYP/CAM- B3LYP87–89 exchange–correlation functional as well as two correlation-consistent basis sets: cc-pVDZ and aug-cc-pVTZ.90 All
electron population descriptors introduced in the text were calcu- Fig. 2 Isosurfaces of ED (blue) and EDDB (green), generated at tb = 0.001 64,65 lated within the NAO-representation obtained from the NBO6 (bonding threshold) and tr = 0.015 (isosurface values), with the corres- d 91 ponding electron populations N (second column), global (black numbers) software by means of several computer scripts originally devel- d and the Kekule´-like (bold burgundy numbers), populations NC (colored oped by the first author; ED and EDDB contour maps were numbers alongside atoms) and natural atomic charges (colored numbers 79 80 obtained using visualization programs Molden and MacMolPlt below molecule) for several simple AHs. Method: CAM-B3LYP/aug-cc- with a number of manually prepared special input files. pVTZ/NAO, equilibrium geometries.
This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 | 20517 View Article Online
PCCP Paper
A more detailed discussion of these points can be found in our previous paper.72 It is worth noticing that, as follows directly from the analysis of along-ring EDDB-populations (bold burgundy numbers inscribed in cyclic structures), resonances between along-ring (Kekule´-like) and cross-ring (Dewar-like) atomic inter- actions are particularly important in charged aromatic hydro- carbons. Preliminary results of more insightful analysis indicate that, contrary to the para-p-delocalization effect observed in the benzene molecule, in other charged AHs also the meta- p-delocalization as well as s-delocalizations (including even hydrogen atoms) play important roles.
3.2 Polycyclic aromatic hydrocarbons
Isosurfaces of EDDB, EDDBp and EDp with the corresponding electron populations for selected polycyclic aromatic hydrocar- bons (Fig. 3 and 4) were calculated using the B3LYP/cc-pVDZ/
NAO method (equilibrium geometries) at tb = 0.001 and tr = 0.015. Black numbers denote global populations while bold burgundy numbers refer to populations of electrons delocalized
Fig. 3 Isosurfaces of EDDB, generated at tb = 0.001 (bonding threshold)
and tr = 0.015 (isosurface values), with the corresponding electron d populations N (first column), global (black numbers) and the Kekule´-like d (bold burgundy numbers), populations NC (second column) for several simple PAHs. Method: B3LYP/cc-pVDZ/NAO, equilibrium geometries.
t Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57. and r = 0.015 (isosurface values). In the last column formal charges of ions (inscribed in molecular rings), total natural atomic charges of all carbon atoms as well as atomic populations of delocalized electrons obtained from eqn (14) are displayed. It should be noticed that, in contrast to the example presented in Fig. 1, Fig. 2, 3 and 5 the sum of all EDDB populations assigned to carbon atoms slightly differs from total EDDB populations reported below each structure. This is mainly due to very minor but notice- able contribution of hydrogen atoms to multicenter bonding (usually 0.01–0.03e per atom). It is evident even from a cursory analysis of numbers in Fig. 2 that, to a greater or lesser extent, populations N d differ + from the expected Hu¨ckel’s numbers: ‘‘2’’ for C3H3 , and 2+ + C4H4 and ‘‘6’’ for C5H5 ,C6H6, and C7H7 . Essentially, there are three reasons for these discrepancies: (1) Nd counts for electrons from both, p- as well as s-delocalization, (2) the cross- ring interactions between carbon atoms and (3) C–H bonds of Fig. 4 (a) Isosurfaces of EDp (left column) and EDDBp, generated at charged AHs are much more polarized revealing tendency to tb = 0.001 (bonding threshold) and tr = 0.010 (isosurface values), with the corresponding electron populations Np and Nd,p and the HMO resonance somewhat overgenerous accumulation of electrons on carbon energies for benzene and several small PAHs. (b) Correlations between p d,p atoms (it follows directly from comparison of total atomic charges resonance energies and the overall populations N (EDp), N (EDDBp)and of all carbon atoms with formal charges of molecules in Fig. 2). Nd (EDDB). Method: B3LYP/cc-pVDZ/NAO, equilibrium geometries.
20518 | Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 This journal is © the Owner Societies 2014 View Article Online
Paper PCCP
only along each cyclic unit. The second column presents atomic An insightful investigation of possible resonance structures d populations Na (colored numbers near atoms); resonance ener- of the coronene molecule indicates that, according to the well- 98,99 gies (Fig. 4) for EDp were calculated at the Hu¨ckel Molecular known Clar’s rule, the center benzenoid is less aromatic Orbital (HMO) theory level.92–95 than external rings. This evidently contradicts the delocaliza- Fig. 3 clearly shows that in several cases even qualitative tion pattern that follows from the EDDB-based population analysis of EDDB contours enables one to predict the relative analysis; the latter accords to some extent with the picture of aromatic stabilization of respective cyclic units that is in coronene proposed by Popov et al.100 As a matter of fact, agreement with the common knowledge about the reactivity analysis of EDDBs of all PAHs larger than naphthalene allows of these species. In particular, it is evident from comparing for the conclusion that the Clar’s rule is fulfilled only for EDDBs and the corresponding global populations for anthra- species having a single unambiguous Clar structure (the same cene and phenanthrene molecules that multicenter electron conclusion has been drawn previously by G. Portella et al.101). delocalization is more effective by about 0.5e in the latter one. Moreover, it follows from a comprehensive analysis of a larger Accordingly, resonance energies (Fig. 4a) for these molecules group of PAHs that at the level of the Hu¨ckel MO method local are 3.60 eV and 3.95 eV, respectively. Quantitative analysis of aromaticities of the overwhelming majority of species satisfy EDDB-based populations of electrons delocalized only along Clar’s rule regardless of the number of equivalent Clar struc- each cyclic unit (bold burgundy numbers) leads to the con- tures per each molecule.22,101 One should realize, however, that clusion that the most highly resonance-stabilized rings are: the HMO method is only a crude approximation that assumes the middle one in anthracene and two side rings in phen- the same idealized geometries for benzenoids and does not anthrene. An in-depth study reveals that 8 atoms in the take into account any s-delocalizations. This, in our opinion, is anthracene terminal rings and 2 atoms in the phenanthrene more than enough to cast doubt on the relative resonance d middle ring have significantly lower values of Na,which stability of cyclic units in coronene predicted by the Clar’s rule. means that their contributions to multicenter bonding in Note that qualitative analysis of the p-layer of electron
both molecules are of minor importance. Consequently, in density (1), EDp, must not necessarily lead to the same conclu-
the anthracene molecule the electrophilic aromatic substitu- sions as the analysis of EDDBp itself. Comparison of EDp and
tion usually involves only the inner ring (two equivalent atoms EDDBp contours for several simple aromatic hydrocarbons d with N a = 1.049) while in the phenanthrene molecule it prefers (Fig. 4a) shows that not the entire p-electron population is d d fouratomsofouterrings(withN a = 0.882 and N a = 0.907). delocalized along aromatic rings, as one might expect. In fact, However, it should be stressed that, in general, local aromati- the effectiveness of p-delocalization varies between 70% (PAHs) city is not simply related to reactivity since the effectiveness of and 92% (benzene). It should be stressed here that this result delocalization in a particular molecular fragment says nothing has been obtained from the first principles and is free from any about the HOMO–LUMO gap or the stability of the transition arbitrariness and references to any idealized system. Moreover, state (e.g., the Diels–Alder reactions always take place in the as follows directly from the analysis of p-electron populations central ring of the anthracene molecule in spite of the fact that and their correlations with the corresponding resonance ener-
it is a more resonance-stabilized cyclic unit than the terminal gies (Fig. 4b), the EDp-populations fail in predicting of relative benzenoids). Nevertheless, the relative aromaticity of the aromatic stabilization of iso-p-electronic systems and only the
Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57. central and external rings in anthracene is still a matter of EDDB- and EDDBp-populations are able to reliably evaluate controversy in the literature.96 global aromaticity of all molecules. Another important conclusion can be drawn from the analysis of the EDDB and the corresponding population 3.3 d-Aromaticity and atypical aromatics numbers referring to electrons delocalized along a particular Multicenter delocalized electron density contours and the ring (bold burgundy numbers) in the case of a fluoranthene corresponding atomic populations can be very helpful in probing molecule (Fig. 3). As follows from these numbers, only the molecular systems with electron delocalization involving d-block benzenoid-like cyclic units are found to be aromatic, and, transition metals. Fig. 5 presents populations and isosurfaces of 2 what is more important, the whole molecule can be regarded EDDBs for the porphine dianion (Por ), the cobalt(II) porphine as built up from the naphthalene unit (3.536e per ring complex (Co-Por) as well as maltol complexes with a vanadyl
compared to 3.529e per ring in a separate naphthalene mole- dication (VO(Ma)2) and aluminium (Al(Ma)3). Calculations were cule) being cross-linked to the benzene unit (4.997e compared performed using the B3LYP/cc-pVDZ/NAO method (equilibrium
to 5.295e in a separate benzene molecule). Indeed, it is well- geometries) at tb = 0.001 and tr = 0.015; the last column presents d known that application of the Hu¨ckel rules sometimes leads atomic populations N a or the total number of electrons deloca- to the conclusion that particular polycyclic compounds (e.g. lized over a particular molecular fragment. Comparing porphine fluoranthene or pyrene) should be anti-aromatic, which dis- with its cobalt complex it is clear that the central atom partici- agrees with their known chemical properties. The examination pates in electron delocalization. Quantitative analysis exhibits of such PAHs as conjugated cyclic polyenes which are intern- some outflow of delocalized electron population from the ring ally cross-linked and/or linked to other cyclic polyenes was due to the presence of the cobalt atom (bridging character of the historically the first commonly accepted solution for this central atom). This back-donation arises only within systems with problem.97 d-electron central atoms. For the next two examples in Fig. 5,
This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 | 20519 View Article Online
PCCP Paper
Fig. 5 Isosurfaces of EDDB, generated at tb = 0.001 (bonding threshold) Fig. 6 Isosurfaces of EDDB, generated at tb = 0.001 (bonding threshold)
and tr = 0.015 (isosurface values), with the corresponding electron and three different isosurface values, tr = 0.010, 0.015, 0.020, respectively, populations Nd (first column) and constituent atomic/fragment popula- with the corresponding electron populations Nd for the homotropylium tions (second column) for porphine dianions, cobalt(II) porphine com- cation and the cyclononatetraenyl cation. Additionally, the most highly
plexes, and vanadyl(IV) and aluminium maltol complexes. Method: B3LYP/ occupied EDDB-derived natural orbitals (generated at tb = 0.001 and tr = cc-pVDZ/NAO, equilibrium geometries. 0.015) and the corresponding occupation numbers are displayed. Method: B3LYP/cc-pVDZ/NAO, equilibrium geometries.
Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57. VO(Ma)2 and Al(Ma)3, the total number of electrons delocalized over the maltol unit is greater in the former case for about 0.2e, In turn, solving the eigenproblem of the EDDB matrix for the which is in agreement with the well-known facts about the cyclononatetraenyl cation gives rise to the well-known Mo¨bius- resonance stabilization of maltol in its chelatoaromatic like orbitals.105 complexes.102,103 Sometimes, especially in the case of non-planar and atypical aromatic molecules, natural orbitals that diagonalize the EDDB 4 Conclusions matrix can give additional insight into the electronic structure of the studied systems. Fig. 6 shows isosurfaces of EDDB, To summarize, there are several important features that set the calculated using the B3LYP/cc-pVDZ/NAO method (equilibrium newly proposed method apart from other measures of multi-
geometries) at three different values of density, tr = 0.010, center delocalization in aromatic rings. (1) Universality – the 0.015, 0.020, with the corresponding electron populations for EDDB-based populations can be easily calculated for planar the homotropylium cation and the cyclononatetraenyl cation. and non-planar molecular rings and therefore they can be Additionally, the most highly occupied natural orbitals of the successfully used for the study of a wide range of aromatic
EDDB matrix selected for both structures are presented (tb = species including both the Hu¨ckel- and Mo¨bius-type aro- 106 104 0.001 and tr = 0.015). The analysis of eigenvectors and eigen- matics, homoaromatics and even non-cyclic aromatic values of the EDDB matrix for the homotropylium cation molecules.107 (2) Intuitiveness and interpretative simplicity – indicates that, besides the evident p-homoconjugation, some quantifying multicenter bonds involves the language of the residual s-delocalization through the methylene carbon atom first-order population analysis. (3) The lack of arbitrariness exists; this is a new result that might have important implica- connected with the necessity of predefining the size of molec- tions for the origin of homoaromatic stabilization effects.104 ular cyclic units (like in the MCI-based techniques) when
20520 | Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 This journal is © the Owner Societies 2014 View Article Online
Paper PCCP
constructing the EDDB. (4) Ability for strict separation of s and 15 H. C. Longuet and J. Higgins, J. Chim. Phys. Phys.-Chim. p components of multicenter delocalization and investigation Biol., 1949, 46, 275. of their mutual interplay. (5) Local, semi-local or global char- 16 A. B. Sannigrahi and T. Kar, Chem. Phys. Lett., 1990, acter of the populations of multicenter delocalized electrons, 173, 569. depending on the strategy of the EDDB matrix construction. 17 R. Ponec and F. Uhlik, Croat. Chem. Acta, 1996, 69, 941. The main purpose of this paper was to introduce a new 18 R. Ponec and D. L. Cooper, Int. J. Quantum Chem., 2004, theoretical approach, describe computational details and 97, 1002. briefly demonstrate its performance on several representative 19 R. Carbo´-Dorca and P. Bultinck, J. Math. Chem., 2004, aromatic species. A comprehensive comparison of EDDB-based 36, 201. delocalization descriptors with a multitude of aromatic stabili- 20 R. Carbo´-Dorca and P. Bultinck, J. Math. Chem., 2004, zation measures based on structural, thermodynamic and 36, 231. magnetic criteria of aromaticity has already been performed 21 M. Giambiagi, M. S. de Giambiagi, C. D. dos Santos Silva and the corresponding paper is currently under preparation. and A. P. de Figueiredo, Phys. Chem. Chem. Phys., 2000, The examples presented in this work show that the electron 2, 3381. density of delocalized bonds is a powerful tool in searching and 22 P. Bultinck, R. Ponec and S. Van Damme, J. Phys. Org. probing electron delocalization in systems of conjugated Chem., 2005, 18, 706. chemical bonds. 23 R. L. Fulton, J. Phys. Chem., 1993, 97, 7516. The EDDB definition introduced in this paper involves the 24 J. G. A´ngya´n, M. Loos and I. Mayer, J. Phys. Chem., 1994, spin-less density matrix and as such it is appropriate for both 98, 5244. closed- and open-shell one-determinant wavefunctions. On the 25 X. Fradera, M. A. Austen and R. F. W. Bader, J. Phys. Chem. basis of several previous investigations,18,39,71 in the nearest A, 1999, 103, 304. future we plan to generalize the method to cover also multi- 26 E. Matito, M. Sola`, P. Salvador and M. Duran, Faraday determinant wavefunctions of both, ground- and excited-state Discuss., 2007, 135, 325. molecular systems. 27 M. Penda´s and F. Blanco, J. Chem. Phys., 2007, 127, 144103. 28 J. Poater, X. Fradera, M. Duran and M. Sola`, Chem. – Eur. J., 2003, 9, 400. Acknowledgements 29 R. Ponec, P. Bultinck and A. G. Saliner, J. Phys. Chem. A, 2005, 109, 6606. This research was supported in part by PL-Grid Infrastructure, 30 M. Mandado, P. Bultinck, M. J. Gonza´lez-Moa and with the calculations performed on Zeus: HP Cluster Platform R. A. Mosquera, Chem. Phys. Lett., 2006, 433,5. of the Academic Computer Centre CYFRONET. 31 M. Mandado, N. Otero and R. A. Mosquera, Tetrahedron, 2006, 62, 12204. 32 F. Feixas, E. Matito, J. Poater and M. Sola`, J. Comput. References Chem., 2008, 29, 1543. 33 M. Sola`, F. Feixas, J. O. C. Jime´nez-Halla, E. Matito and
Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57. 1 G. N. Lewis, J. Am. Chem. Soc., 1916, 38, 762. J. Poater, Symmetry, 2010, 2, 1156. 2 I. Langmuir, J. Am. Chem. Soc., 1919, 41, 868. 34 R. F. Nalewajski, Information theory of molecular systems, 3 R. S. Mulliken, Rev. Mod. Phys., 1932, 4,1. Elsevier, Amsterdam, 2006. 4 S. F. Boys, Rev. Mod. Phys., 1960, 32, 296. 35 R. F. Nalewajski, Information origins of the chemical bond, 5 C. Edmiston and K. Ruedenberg, J. Chem. Phys., 1965, Nova Science Publishers, Hauppauge, NY, 2010. 43, 97. 36 R. F. Nalewajski, D. W. Szczepanik and J. Mrozek, 6 J. Pipek and P. G. Mezey, J. Chem. Phys., 1989, 90, 4916. Adv. Quantum Chem., 2011, 61,1. 7 A. E. Reed, L. A. Curtiss and F. Weinhold, Chem. Rev., 1988, 37 D. W. Szczepanik and J. Mrozek, J. Math. Chem., 2011, 88, 899. 49, 562. 8 K. Jug, J. Am. Chem. Soc., 1977, 99, 7800. 38 R. F. Nalewajski, D. W. Szczepanik and J. Mrozek, J. Math. 9 A. Michalak, M. Mitoraj and T. Ziegler, J. Phys. Chem., 2008, Chem., 2012, 50, 1437. 112, 1933. 39 D. W. Szczepanik and J. Mrozek, J. Math. Chem., 2013, 10 W. N. Lipscomb, Boron Hydride Chemistry, Academic Press, 51, 1388. New York, 1975. 40 D. W. Szczepanik and J. Mrozek, J. Math. Chem., 2013, 11 R. E. Dickerson and W. N. Lipscomb, J. Chem. Phys., 1957, 51, 1619. 27, 212. 41 J. Cioslowski, E. Matito and M. Sola`, J. Phys. Chem. A, 2007, 12 W. N. Lipscomb, Acc. Chem. Res., 1973, 8, 257. 111, 6521. 13 L. Pauling, The Nature of the Chemical Bond, Cornell 42 R. S. Mulliken, J. Chem. Phys., 1955, 23, 1833. University Press, New York, 1960. 43 R. S. Mulliken, J. Chem. Phys., 1955, 23, 1841. 14 K. Wade, Electron Deficient Compounds, Studies in Modern 44 R. S. Mulliken, J. Chem. Phys., 1955, 23, 2338. Chemistry, Nelson, London, 1971. 45 R. S. Mulliken, J. Chem. Phys., 1955, 23, 2343.
This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 | 20521 View Article Online
PCCP Paper
46 A. J. Bridgeman and C. J. Empson, New J. Chem., 2008, 77 M. S. Gordon and M. W. Schmidt, Advances in electronic 32, 1359. structure theory: GAMESS a decade later, in Theory and 47 M. S. de Giambiagi, M. Giambiagi and M. de Souza Fortes, Applications of Computational Chemistry: the First Forty THEOCHEM, 1997, 391, 141. Years, ed. C. E. Dykstra, G. Frenking, K. S. Kim and 48 A. S. Mikhaylushkin, J. Nyle´n and U. Ha¨ussermann, Chem. G. E. Scuseria, Elsevier, Amsterdam, 2005, pp. 1167–1189. – Eur. J., 2005, 11, 4912. 78 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, 49 A. B. Sannigrahi, P. K. Nandi, L. Behara and T. Kar, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, THEOCHEM, 1992, 95, 259. B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, 50 C. G. Bollini, M. Giambiagi, M. S. de Giambiagi and X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, A. P. de Figueiredo, Struct. Chem., 2001, 12, 113. J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, 51 J. P. Foster and F. Weinhold, J. Am. Chem. Soc., 1980, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, 102, 7211. H. Nakai, T. Vreven, J. A. Montgomery, J. E. Peralta, 52 J. E. Carpenter and F. Weinhold, THEOCHEM, 1988, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, 169, 41. K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, 53 F. Feixas, E. Matito, M. Duran, J. Poater and M. Sola`, Theor. K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, Chem. Acc., 2011, 128, 419. J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, 54 F. Weinhold and C. R. Landis, Valency and Bonding: A J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, Natural Bond Orbital Donor-Acceptor Perspective, Cambridge R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, University Press, Cambridge, 2005. R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, 55 F. Weinhold and C. R. Landis, Discovering Chemistry with K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, Natural Bond Orbitals, Wiley, Hoboken, NJ, 2012. J. J. Dannenberg, S. Dapprich, A. D. Daniels, O¨. Farkas, 56 D. Y. Zubarev and A. I. Boldyrev, Phys. Chem. Chem. Phys., J. B. Foresman, J. V. Ortiz, J. Cioslowski and D. J. Fox, 2008, 10, 5207. Gaussian 09, Revision D.01, Gaussian, Inc., Wallingford CT, 57 R. F. Nalewajski, J. Math. Chem., 2011, 49, 371. 2009. 58 R. F. Nalewajski, J. Math. Chem., 2011, 49, 546. 79 G. Schaftenaar and J. H. Noordik, J. Comput.-Aided Mol. 59 R. F. Nalewajski, J. Math. Chem., 2011, 49, 1226. Des., 2000, 14, 123. 60 R. F. Nalewajski, Int. J. Quantum Chem., 2012, 112, 2355. 80 B. M. Bode and M. S. Gordon, J. Mol. Graphics Modell., 61 R. F. Nalewajski, Int. J. Quantum Chem., 2013, 113, 766. 1998, 16, 133. 62 D. W. Szczepanik and J. Mrozek, Comput. Theor. Chem., 81 D. W. Szczepanik and J. Mrozek, Chem. Phys. Lett., 2012, 2013, 1026, 72. 521, 157. 63 U. Fano, Rev. Mod. Phys., 1957, 29, 74. 82 D. W. Szczepanik and J. Mrozek, Comput. Theor. Chem., 64 A. E. Reed and F. Weinhold, J. Chem. Phys., 1983, 78, 2013, 1008, 15. 4066. 83 D. W. Szczepanik and J. Mrozek, Comput. Theor. Chem., 65 A. E. Reed, R. B. Weinstock and F. Weinhold, J. Phys. 2012, 996, 103. Chem., 1985, 83, 735. 84 R. S. Mulliken and W. C. Ermler, Diatomic Molecules:
Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57. 66 K. Wiberg, Tetrahedron, 1968, 24, 1083. Results of Ab Initio Calculations, Academic Press, New York, 67 L. Pauling, J. Am. Chem. Soc., 1947, 69, 542. 1977. 68 N. P. Borisova and S. G. Semenov, Vestn. Leningr. Univ., 85 J. B. Collins and A. Streitwieser, J. Comput. Chem., 1980, 1973, 16, 119. 1, 81. 69 I. Mayer, J. Comput. Chem., 2007, 28, 204. 86 H. P. Lu¨thi, J. H. Ammeter, J. Almlo¨f and K. Faegri, J. Chem. 70 D. W. Szczepanik and J. Mrozek, J. Theor. Comput. Chem., Phys., 1982, 77, 2002. 2011, 10, 471. 87 K. Kim and K. D. Jordan, J. Phys. Chem., 1994, 98, 71 D. W. Szczepanik and J. Mrozek, Comput. Theor. Chem., 10089. 2013, 1023, 83. 88 P. J. Stephens, F. J. Devlin, C. F. Chabalowski and M. J. 72 D. W. Szczepanik, E. Z˙ak, K. Dyduch and J. Mrozek, Chem. Frisch, J. Phys. Chem., 1994, 98, 11623. Phys. Lett., 2014, 593, 154. 89 T. Yanaia, D. P. Tewb and N. C. Handyb, Chem. Phys. Lett., 73 J. Poater, X. Fradera, M. Duran and M. Sola`, Chem. – Eur. J., 2004, 393, 51. 2003, 9, 1113. 90 T. H. Dunning, J. Phys. Chem., 1989, 90, 1007. 74 E. Matito, P. Salvador, M. Duran and M. Sola`, J. Phys. Chem. 91 E. D. Glendening, C. R. Landis and F. Weinhold, J. Comput. A, 2006, 110, 5108. Chem., 2013, 34, 1429. 75 M. Gu¨ell, E. Matito, J. M. Luis, J. Poater and M. Sola`, 92 E. Hu¨ckel, Z. Phys., 1931, 70, 104. J. Phys. Chem. A, 2006, 110, 11569. 93 E. Hu¨ckel, Z. Phys., 1931, 72, 310. 76 M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T. Elbert, 94 E. Hu¨ckel, Z. Phys., 1932, 76, 628. M. S. Gordon, J. H. Jensen, S. Koseki, N. Matsunaga, 95 E. Hu¨ckel, Z. Elektrochem., 1937, 43, 752. K. A. Nguyen, S. Su, T. L. Windus, M. Dupuis and 96 G. Portella, J. Poater, J. M. Bofill, P. Alemany and M. Sola`, J. A. Montgomery, J. Comput. Chem., 1993, 14, 1347. J. Org. Chem., 2005, 70, 2509; erratum: G. Portella, J. Poater,
20522 | Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 This journal is © the Owner Societies 2014 View Article Online
Paper PCCP
J. M. Bofill, P. Alemany and M. Sola`, J. Org. Chem., 2005, 102 K. K. Zborowski, R. Grybos´ and L. M. Proniewicz, Inorg. 70, 4560. Chem. Commun., 2005, 8, 76. 97 A. Streitwieser, Molecular Orbital Theory for Organic 103 K. K. Zborowski, M. Sola`, J. Poater and L. M. Proniewicz, Chemists, Wiley, New York, 1961. J. Phys. Org. Chem., 2011, 24, 499. 98 E. Clar, Polycyclic hydrocarbons, Academic Press, London/ 104 S. Winstein, J. Am. Chem. Soc., 1959, 81, 6523. New York, 1964. 105 H. S. Rzepa and N. Sanderson, Phys. Chem. Chem. Phys., 99 E. Clar, The Aromatic Sextet, Wiley, New York, 1972. 2004, 6, 310. 100 I. A. Popov and A. I. Boldyrev, Eur. J. Org. Chem., 2012, 3485. 106 H. E. Zimmerman, J. Am. Chem. Soc., 1966, 88, 1564. 101 G. Portella, J. Poater and M. Sola`, J. Phys. Org. Chem., 2005, 107 A. Dworkin, R. Naumann, C. Seigfred and J. M. Karty, 18, 785. J. Org. Chem., 2005, 70, 7605. Published on 11 August 2014. Downloaded by UNIWERSYTET JAGIELLONSKI 24/09/2014 13:56:57.
This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys., 2014, 16, 20514--20523 | 20523 Computational and Theoretical Chemistry 1080 (2016) 33–37
Contents lists available at ScienceDirect
Computational and Theoretical Chemistry
journal homepage: www.elsevier.com/locate/comptc
A new perspective on quantifying electron localization and delocalization in molecular systems ⇑ Dariusz W. Szczepanik
K. Gumin´ski Department of Theoretical Chemistry, Faculty of Chemistry, Jagiellonian University, Ingerdena 3, 30-060 Cracow, Poland article info abstract
Article history: The original method of electron density partitioning is introduced that allows one to probe electron local- Received 15 January 2016 ization and delocalization within one theoretical paradigm. The newly proposed method makes use of the Received in revised form 29 January 2016 age-old concept of bond-order orbitals as well as the recently developed bond-orbital projection formal- Accepted 2 February 2016 ism to decompose the one-electron density into density layers representing electrons localized on atoms Available online 8 February 2016 (inner shells, lone pairs), shared between atoms (chemical bonds) and delocalized between adjacent bonds (multi-center bonding). The details of the current implementation are briefly discussed and several Keywords: illustrative examples are provided. Electron localization Ó 2016 Elsevier B.V. All rights reserved. Electron delocalization Density partitioning Aromaticity
Our understanding of the chemical structure and reactivity is The definition of the EDDAðrÞ component is crucial for the entire usually built up from and dependent upon such intuitive concepts method. In contrary to other approaches regarding the electron as atom in molecule, chemical bond, lone pair, Lewis structure, density/population distributions within physical space [7–9], here bond covalency and ionicity, etc. [1,2]. Although the semantics of we use the Hilbert-space partitioning scheme based on the one- these fundamental entities is not sharply defined in modern quan- electron density matrix,noD, within the basis of the natural atomic tum mechanics, a multitude of different formalisms has been pro- orbitals (NAO) [10], vlðrÞ , but any other set of orthonormal and posed in the literature either to determine them as functions of the still well atom-assigned functions can also be used [11,12]. Here, electron density (ED), like within the framework of the quantum it has to be noticed that the presented formalism is restricted to theory of atoms in molecules (QTAIM) [3,4], or provide them a one-determinant wave functions only and thus it can be used at localized-orbital representation, like the natural bond orbital HF and DFT theory levels (bear in mind, however, that the Kohn– (NBO) approach does [5]. Despite great popularity and success, Sham determinant relies on the auxiliary set of orbitals represent- both these approaches have certain drawbacks that sometimes ing non-interacting electrons and as such it provides only an may cut back their applicability, eg. the shortcomings regarding approximation to the first order density matrix). The definitions criteria for the chemical bond to exist [6], extremely high compu- given in the text are correct for closed-shell systems. The spin- tational cost of the localized-orbital description for large molecules resolved calculations for open-shell systems should be carried out with multi-center bonding, and many others. In this paper, for the separately for both a- and b-spin density matrices and then, in first time these two methodologies are unified to provide a scheme accordance with the Gopinatham-Jug definition of the bond- for the exact decomposition of the one-electron density into the covalency index [13], the resulting sum of spin-components needs density layers representing electrons well-localized on atoms to be multiplied by 2. (EDLA), like core electrons, lone pairs and ionic bonds, and the elec- The construction of the EDDAðrÞ function requires solving a set trons delocalized between all atoms in a molecule, (EDDA), ie. of the eigenproblems for all possible Jug’s matrices in the molecu- covalent bonds, lar system [14]. The classical Jug’s matrix is determined by the EDðrÞ¼EDLAðrÞþEDDAðrÞ: ð1Þ a; b off-diagonal ( )-diatomic blocks of the one-electron density matrix ( Dab ) [15], ! ⇑ 0Da;b y Tel.: +48 12 663 20 25. y ¼ CabkabCab !jfabi¼jviCab; ð2Þ E-mail address: [email protected] Da;b 0 URL: http://www.chemia.uj.edu.pl/~szczepad http://dx.doi.org/10.1016/j.comptc.2016.02.003 2210-271X/Ó 2016 Elsevier B.V. All rights reserved. 34 D.W. Szczepanik / Computational and Theoretical Chemistry 1080 (2016) 33–37 no no X X DA DA b n N ¼ aDDA ¼ N : ð Þ and its spectrum contains positive ( ka;b ), zero ( ka;b ) and nega- tr a 9 no a a tive eigenvalues ( ka ), associated with the corresponding eigen- a;b In accordance with Eq. (1), the density layer representing elec- b n vectors representing bonding (jfabi), non-bonding (jfabi) and anti- trons well-localized on atomic centers, EDLA(r), can be straightfor- a bonding (jfabi) two-center bond orbitals (2cBO), respectively, i.e. wardly obtain by the subtraction of EDDA(r) from the one-electron density, ED(r). But, furthermore, as shown very recently by the b n a b n a kab ¼ kabjkabjkab ; Cab ¼ CabjCabjCab : ð3Þ author [14,19], the EDDA(r) layer can be regarded as a sum of two components representing electron density of localized and As originally pointed out by Jug [15], the sum of squared elements delocalized bonds, EDLB(r) and EDDB(r), respectively. The latter b of the diagonal matrix kab is identical to the bond-order definition is determined through the bond-orbital projection technique, by Wiberg [16], Wab, which, alongside involving the subset of appropriately orthogonal- a b ized [20] 2cBOs, requires also their three-center counterparts [21] 2 X X ¼ kb ¼ 1 2 : ð Þ to represent all the possible bond-conjugations in the molecular Wab tr ab Dab l;m 4 2 l m system. Thus, taking into account additional step of the EDDA(r) This property (characteristic only for idempotent density matrices decomposition, the electron density can be partitioned as follows: [17]) is used to define the quasi-metric of all bonding 2cBOs within the representation of NAOs, EDðrÞ¼EDLAðrÞþEDLBðrÞþEDDBðrÞ: ð10Þ X 2 y To illustrate how the newly proposed ED-partitioning proce- Sb ¼ Cb kb Cb ; ð5Þ f ab ab ab dure works in practice, it has been used to probe the electronic a;b structure of the valence shells of benzene, s-triazine and borazine, When the summation in (5) includes only atomic pairs that repre- calculated at the B3LYP/6-311++G⁄⁄ theory level, as well as the ⁄⁄ b sents typical covalent interactions then the trace of the Sf matrix anthracene molecule calculated using the CAM-B3LYP/6-311+G can be regarded as a good approximation to the population of all method. The quantum-chemical software including Firefly [22], shared-electron pairs in the molecular system (i.e. the population MultiWFN [23] and Molden [24] has been used to perform compu- ð Þ of electrons delocalized through the net of all chemical bonds). tations and visualize the results. The EDDB r density layers have r p Technically, this is the case only if the highest occupied 2cBOs form been dissected into and components to provide a more detail description of the bond-conjugation effects [18]. a subset of the NBO-like orthonormalno orbitals associated with the The results, as presented in Fig. 1, clearly shows that the nearly degenerated eigenvalues ðkb Þ . However, if one takes ab l;l EDLAðrÞ function easily copes with detection of lone-pairs in into account also the weak non-covalent interactions or if the s-triazine as well as regions of highly localized electrons due to multi-center bonding occurs, the 2cBOs cannot be expanded as lin- charge polarization (borazine, but also to a small extent the ben- ear combinations of the doubly-occupied molecular orbitals (MO) zene molecule). However, in contrast to the expectations depen- b only. To solve this problem and give the diagonal of the Sf matrix dent upon the qualitative model of bonding by Lewis, in the case a strict populational character, it is proposed to project the eigen- of borazine the six p-electrons do not form a typical aromatic ring b but tend to remain as unhybridized electron-pairs centered on vectors of Sf onto the ground-state molecular wave function or, which is fully equivalent, transform the quasi-metric (5) under nitrogen atoms. This fact explains remarkable electron localization the following orthogonal similarity transformation: on nitrogen atoms, but also rationalizes significantly less effective multi-center delocalization quantified by the EDDBp component, ~ b 1 y b b ~ b Sf ¼ 2 D Sf D; N ¼ tr½Sf : ð6Þ which is in full agreement with findings by other researchers [25]. In turn, the effectiveness of p-delocalization in both benzene Here, one should realize that the total population of electrons delo- and s-triazine is very high, but still slightly lower than expected for b calized between atoms in a molecule, N , is markedly underesti- an ideal aromatic 6p-electron system. Here, the reason is that the mated if the bond-conjugation effect occurs [18]. This is especially cross-ring interactions, which are particularly important for important for aromatic species, in which the shared p-electron pairs para-related atoms, are reluctant to conjugate with the covalent are delocalized between more than two atomic centers and thus bonds along molecular ring, remaining weak but well-localized they cannot be exactly described by any subset of localized bond ‘bonds’ in nature. Although this fact has already been noticed in orbitals. It should be stressed, however, that in such cases the the literature [19], no comprehensive research has been reported bond-order representability condition for the matrix (5) determines so far regarding the influence of cross-ring interactions on the significant contribution of the virtual MOs to the shape of bonding effectiveness of bond-conjugation in aromatics. Here, it should be 2cBOs (this property of the 2cBOs will be the subject of a separate mentioned that the para-delocalization effect is the basis of the study). If so, the complementary set of anti-bonding 2cBOs can be very popular aromaticity index by Poater et.al. (PDI) [26]. Since it used to ‘recover’ the lacking electron population in (6). Therefore, has been reported that in some cases this descriptor is not suitable in analogy to (5) and (6), the GS-projected quasi-metric of all for the description of local aromaticity due to overestimation of the anti-bonding 2cBOs within the representation of NAOs is given by "# contribution of Dewar’s resonance structure to the electron delo- X 2 calization [27], a comparative study of PDI with EDLB and EDDB S~ a ¼ 1 y Ca ka Cay : ð Þ f 2 D ab ab ab D 7 components of the electron density seem to be even more strongly a;b justified. In contrast to the p-type bonds, which have a much Then, the entrywise sum of the matrices (6) and (7) gives rise to the higher tendency to conjugation in the unsaturated species, the NAO-based density matrix of the EDDAðrÞ function, and hence r-bonding represents rather typical localized two-center bonds X in all three molecular systems. Again, the EDDBrðrÞ function reaf- ð Þ¼ vy ð ÞDDA v ð Þ; DDA ¼ S~ b þ S~ a; ð Þ EDDA r l r l;m m r f f 8 firms this fact setting the magnitude of r-delocalization up to l;m about 5% of total population of r-electrons. The corresponding total population of electrons delocalized The newly proposed ED-partitioning scheme can also be very between atoms in a molecule can be easily partitioned into atomic useful in rationalizing the electronic structure of polycyclic sys- contributions: tems with conjugated p-bonds. In particular, it is well-known that D.W. Szczepanik / Computational and Theoretical Chemistry 1080 (2016) 33–37 35
Fig. 1. Contours of the valence-shell electron density and its components visualized at isovalue s ¼ 0:015, and the corresponding electron populations obtained by integration of the appropriate function over the whole space. Method: B3LYP/6-311++G⁄⁄ (equilibrium geometries). the reactivity of anthracene is determined by its electron delocal- neglect or underestimate the bond-bond delocalization effects. ization pattern in the transition state, predicting the external rings Indeed, as depicted in Fig. 2, the cyclic delocalization patterns in to be more aromatic [28]. However, in the ground state the picture anthracene provided by such popular tools as the electron localiza- of p-delocalization in anthracene is not clear and only a few aro- tion function (ELF) [30], localized orbital localizator (LOL) [31] or maticity quantifiers support the finding based on the structural the laplasian of the electron density, seem to be topologically and magnetic criteria of aromaticity that the p-conjugation is more closer to the EDLB function. If one takes into account that the effective in the central ring [29]. It has recently been pointed out EDLB-component represents only about 35% of the total that the ‘‘anthracene problem” introduced by the electronic criteria p-electron population in anthracene, while the EDDB population of aromaticity may be a consequence of using quantities and tools is nearly twice bigger, it become quite obvious that the EDDB that favors the effect of electron delocalization between atoms and component is dominating and determines the local aromaticity of 36 D.W. Szczepanik / Computational and Theoretical Chemistry 1080 (2016) 33–37
functions, but requires the use of additional data files generated by the NBO software [32] and containing the matrix of linear coefficients for MOs in the NAO basis, as well as the AO ! NAO transformation matrix. By default, the program decomposes the one-electron density regarding delocalization effects between all atoms in a molecule. However, if required, one can specify partic- ular molecular fragment, and even select individual bonds that the ED-partitioning should be restricted to. The foregoing results seem to be promising enough to legitimize further and more extensive studies, especially in the field of chemical aromaticity in excited states [33].
Acknowledgments
The author is very grateful to Prof. Tadeusz M. Krygowski, Prof. Miquel Solà and Dr. Marcin Andrzejak for multiple stimulating dis- cussions. This research was supported in part by the Foundation for Polish Science (FNP START 2015 stipend), National Science Centre, Poland (NCN, Grant No. 2015/17/D/ST4/00558), as well as the PL- Grid Infrastructure, with the calculations performed on Zeus: HP Cluster Platform of the Academic Computer Centre CYFRONET.
References
[1] G.N. Lewis, Valence and the Structure of Atoms and Molecules, Chemical Catalog, New York, 1923. [2] D. Szczepanik, J. Mrozek, Probing the interplay between multiplicity and ionicity of the chemical bond, J. Theor. Comput. Chem. 10 (2011) 471–482. [3] R.F.W. Bader, Atoms in Molecules: A Quantum Theory, Oxford University Press, Oxford, 1994. [4] J. Andrés, L. Gracia, P. González-Navarrete, V.S. Safont, Chemical structure and reactivity by means of quantum chemical topology analysis, Comput. Theor. Chem. 1053 (2015) 17–30. [5] A.E. Reed, L.A. Curtiss, F. Weinhold, Intermolecular interactions from a natural bond orbital, donor-acceptor viewpoint, Chem. Rev. 88 (1988) 899–926. [6] J.R. Lane, J. Contreras-García, J.P. Piquemal, B.J. Miller, H.G. Kjaergaard, Are bond critical points really critical for hydrogen bonding?, J Chem. Theory Comput. 9 (2013) 3263–3266. [7] E. Francisco, A.M. Pendás, A. Costales, M. García-Revilla, Electron number distribution functions with iterative Hirshfeld atoms, Comput. Theor. Chem. 975 (2011) 2–8. [8] Z.A.M. Zielinski, Jason K. Pearson, The localized pair model of electronic structure analysis, Comput. Theor. Chem. 1003 (2013) 79–90. [9] A. Morgenstern, T. Wilson, J. Miorelli, T. Jones, M.E. Eberhart, In search of an intrinsic chemical bond, Comput. Theor. Chem. 1053 (2015) 31–37. [10] A.E. Reed, F. Weinhold, Natural bond orbital analysis of near-Hartree-Fock water dimer, J. Chem. Phys. 78 (1983) 4066–4073. Fig. 2. Comparison of p-electron density components EDLB and EDDB with ELF [11] D.W. Szczepanik, J. Mrozek, Electron population analysis using a reference [30], LOL [31] and ED-laplacian function in the anthracene molecule (the minimal set of atomic orbitals, Comput. Theor. Chem. 996 (2012) 103–109. corresponding surface isovalues are collected in the last column). Method: B3LYP/ [12] D.W. Szczepanik, J. Mrozek, Minimal set of molecule-adapted atomic orbitals 6-311+G⁄⁄ (equilibrium geometries). from maximum overlap criterion, J. Math. Chem. 51 (2013) 2687–2698. [13] M.S. Gopinathan, K. Jug, Valency. I. A quantum chemical definition and properties, Theoret. Chim. Acta 63 (1983) 497–509. _ [14] D.W. Szczepanik, M. Andrzejak, K. Dyduch, E. Zak, M. Makowski, G. Mazur, J. antracene to be in line with magnetic and structural criteria of Mrozek, A uniform approach to the description of multicenter bonding, Phys. Chem. Chem. Phys. 16 (2014) 20514–20523. aromaticity. It should also be emphasized that the EDDB [15] K. Jug, A maximum bond order principle, J. Am. Chem. Soc. 99 (1977) 7800– populations have recently been found to be well correlated with 7805. energetic effects of global aromatic stabilization in polycyclic [16] K.B. Wiberg, Application of the Pople–Santry–Segal CNDO method to the Cyclopropylcarbinyl and Cyclobutyl Cation and to Bicyclobutane, Tetrahedron hydrocarbons [18]. 24 (1968) 1083–1096. It is worth to be mentioned that determination of all [17] D.W. Szczepanik, J. Mrozek, Stationarity of electron distribution in ground- ED-components within the proposed method does not entail con- state molecular systems, J. Math. Chem. 51 (2013) 1388–1396. [18] D.W. Szczepanik, Evaluation of aromaticity using the EDDB method (2015) siderable computational costs. On the contrary. The slowest step http://dx.doi.org/10.13140/RG.2.1.4376.5921. Available online:
[25] A.K. Phukan, A.K. Guha, B. Silvi, Is delocalization a prerequisite for stability of [30] A.D. Becke, K.E. Edgecombe, A simple measure of electron localizationh in ring systems? A case study of some inorganic rings, Dalton Trans. 39 (2010) atomic and molecular systems, J. Chem. Phys. 92 (1990) 5397–5403. 4126–4137. [31] H. Jacobsen, Localized-orbital locator (LOL) profiles of chemical bonding, [26] J. Poater, X. Fradera, M. Duran, M. Solà, An insight into the local aromaticities Canad. J. Chem. 86 (2008) 695–702. of polycyclic aromatic hydrocarbons and fullerenes, Chem. Eur. J. 9 (2003) [32] E.D. Glendening, C.R. Landis, F. Weinhold, NBO 6.0: natural bond-orbital 1113–1122. analysis program, J. Comp. Chem. 34 (2013) 1429–1437. [27] P. Bultinck, R. Ponec, Sofie Van Damme, Multicenter bond indices as a new [33] M. Andrzejak, D.W. Szczepanik, Ł. Orzeł, The lowest triplet states of bridged measure of aromaticity in polycyclic aromatic hydrocarbons, J. Phys. Org. cis-2,20-bithiophenes – theory vs experiment, Phys. Chem. Chem. Phys. 17 Chem. 18 (2005) 706–718. (2015) 5328–5337. [28] P.R. Schleyer, J.I. Wu, F.P. Cossio, I. Fernandez, Aromaticity in transition structures, Chem. Soc. Rev. 43 (2014) 4909–4921. [29] P.R. Schleyer, M. Manoharan, H. Jiao, F. Stahl, The acenes: is there a relationship between aromatic stabilization and reactivity?, Org Lett. 3 (2001) 3643–3646. Computational and Theoretical Chemistry 1100 (2017) 13–17
Contents lists available at ScienceDirect
Computational and Theoretical Chemistry
journal homepage: www.elsevier.com/locate/comptc
On the three-center orbital projection formalism within the electron density of delocalized bonds method
Dariusz W. Szczepanik
K. Gumin´ski Department of Theoretical Chemistry, Faculty of Chemistry, Jagiellonian University, Ingerdena 3, 30-060 Cracow, Poland article info abstract
Article history: A new development of the Electron Density of Delocalized Bonds formalism (EDDB) is proposed that pro- Received 21 November 2016 vides marked improvement in the description of electron delocalization in aromatic rings. Special atten- Received in revised form 1 December 2016 tion is paid to charged aromatic hydrocarbons of different size, for which the total population of electrons Accepted 1 December 2016 delocalized between adjacent bonds from the original formulation of the EDDB method significantly Available online 2 December 2016 overestimates the multicenter p-electron sharing effects. The revised bond-orbital projecting scheme gives rise to systematic improvement of the results of the EDDB analysis, which now supports findings Keywords: by other researchers. Bond conjugation Ó 2016 Elsevier B.V. All rights reserved. Orbital projection Electron delocalization EDDB Aromaticity
1. Introduction quantities can be easily calculated for both planar and non- planar molecular rings and do not suffer from the ring-size exten- The electron density of delocalized bonds (EDDB) [1,2] formal- sivity issues. Therefore, they can be successfully used for investiga- ism has recently been proposed for comprehensive description of tions of a wide range of aromatic species. It should be noted in electron delocalization in molecules with conjugated bonds. The passing that, apart from some very specific situations (e.g. [13]), EDDB approach provides both the overall picture of multicenter the EDDB scheme is able to strictly separate r; p and higher com- electron delocalization for the entire molecule and the detailed ponents of multicenter delocalization (even in the case of non- description of the delocalized electron populations in atomic reso- planar ring units) as well as probe their mutual interplay. lution for particular molecular fragment. Visualization of the elec- The EDDB formalism is rooted in the Molecular Orbital (MO) tron density of delocalized bonds facilitates quick detection of theory and, just like other familiar techniques, e.g. the Natural regions of increased aromaticity in very large systems, while the Bond Order (NBO) [14–16] or the Adaptive Natural Density Parti- detailed information on the delocalized populations enables quan- tioning (AdNDP) [17], it is based on the bond-order orbitals formal- titative studies of bond conjugation in a selected molecular frag- ism, which was originally introduced by Jug [18]. In contrast to ment [1–5]. other methods, however, the electron density of delocalized bonds The most important features setting the EDDB analysis apart is constructed only from two-atomic and three-atomic bond-order from other techniques of evaluation of multicenter delocalization orbitals, which considerably reduce computational cost and sim- are the non-referential character, intuitiveness and interpretative plifies the analysis of multicenter bonds in large molecular simplicity. Indeed, unlike with other quantum-chemical measures systems. of the multicenter electron sharing effects [6–10], the electron The originally proposed formulation of the EDDB analysis is population determined by the trace of the EDDB density matrix based on the approximate relationship between two- and three- represents a total number of electrons delocalized through the sys- center Bonding Orbitals (2cBO and 3cBO, respectively), which has tem of conjugated bonds [2]. Depending on the strategy chosen to recently been studied using some information-theoretic tools construct the electron density matrix of delocalized bonds, it can [19–23]. This approximation, however, may sometimes lead to possess local (only conjugations of selected bonds) or global (all quantitatively incorrect predictions of the number of electrons possible bond conjugations in a molecule) character. Also, unlike delocalized in molecular systems. A detailed investigation has other aromaticity descriptors (e.g. NICS [11,12]), the EDDB-based revealed that in some cases, especially for charged aromatic hydro- carbons (AH), the original EDDB scheme systematically overesti- mates the number of electrons assigned to conjugated bonds of E-mail address: [email protected] p URL: http://www.chemia.uj.edu.pl/~szczepad -type [4]. It is the main objective of this paper to introduce a http://dx.doi.org/10.1016/j.comptc.2016.12.003 2210-271X/Ó 2016 Elsevier B.V. All rights reserved. 14 D.W. Szczepanik / Computational and Theoretical Chemistry 1100 (2017) 13–17 modification to the orbital-projecting procedure that remedies this where, by definition, the corresponding LC coefficients read: DE DE issue and improves the EDDB analysis. Tab ¼ ~fb fb ; Tbc ¼ ~fb fb ð Þ abc ab;i abc;m abc bc;j abc;m 8 i;m j;m 2. Theoretical background In the crucial step of the EDDB matrix construction, the coefficients from (8) are used to determine the population of electrons origi- The electron density of delocalized bonds can be straightfor- nally assigned to particular chemical bond, e.g. Xa–Xb, and delocal- wardly extracted from a one-determinant wavefunction of molec- ized through the sequence of bonded atoms Xa–Xb–Xc: ular system obtained by means of standard methods of quantum X 2 2 chemistry. It involves the use of well atom-assigned basis func- d ab d Nabjabc ¼ Tabc kabc : ð9Þ tions, which can be the Natural Atomic Orbitals (NAO) [24,25], i;m m;m no i;m vlðrÞ , or any other representation including the effective d Here, the diagonal matrix kabc collects populations of multicenter minimal-basis subspace of orthonormal atomic orbitals, e.g. [26– delocalized electrons assigned to each 3cBO; more details about 33]. Then, the one-electron EDDB function can be expressed as d XX the definition of kabc can be found in [5]. ð Þ¼ v ð ÞyDd v ð Þ; Nd ¼ Dd: ð Þ EDDB r l r l;m m r tr 1 As indicated in the preceding section, the problem with the l m 3cBO ! 2cBO projection scheme (3)–(9) lies in the fact that it com- Here, Nd stands for the total population of electrons delocalized in pletely neglects the through-space (direct) interaction between d terminal atoms in a system of two conjugated bonds. A simple all conjugated bonds in a molecule while the density matrix Dl;m is solution to this issue is to define the three-center bonding orbitals defined as a sum of ‘‘density layers” representing all possible inter- by solving the eigenproblem of the 3 3 Jug’s matrix that implic- actions between atoms in pairs: XX itly includes the interaction between Xa and Xc, i.e. d d 0 1 D ¼ Bab; ð2Þ 0Da;b Da;c a b–a B C E B y C b Da;b 0Db;c ! n : ð10Þ Bd @ A abc where ab represents this part of the electron density of the chem- y y Da;c Db;c 0 ical bond Xa–Xb that is delocalized in a multicenter sense with all Bd other bonds in molecular system. The construction of ab is a com- For the resulting three-center bonding orbitals, however, the plex task falling outside the framework of this work and it has been appropriately orthogonalized two-center bonding orbitals (5) do comprehensively discussed in two recent papers [2,5]. What should not form a complete basis set. Therefore, the improved be emphasized at this point is that the determination of each layer 3cBO ! 2cBO projection scheme must involve an additional subset of the density E matrix (2) involves the use of three-center bonding of orbitals obtained by solving the eigenproblem of the Jug’s matrix orbitals fb , representing the Xa–Xc bonding: abc ! E E 0Da;c b b ! ~fb : ð Þ fabc ¼jviCabc; ð3Þ y ac 11 Da;c 0 obtained by solving the eigenproblem of the following matrix, 0 1 Then, according to Eq. (6), each 3cBO can be regarded as a linear 0Da;b 0 combination (LC) of orthogonalized 2cBO from three different B C B y C y bonds (either covalent or noncovalent), Xa–Xb,Xb–Xc and Xa–Xc, Dabc ¼ @ Da;b 0Db;c A ¼ CabckabcCabc; ð4Þ respectively, y E E E 0Db;c 0 X X nb ¼ ~fb ab þ ~fb bc abc;m ab;i Tabc bc;j Tabc i;m j;m and selecting the subspace of eigenvectors associated only with i j X E positive eigenvalues [2]; here the elements of the matrix Dabc rep- þ ~fb ac ; ð Þ ac;j Tabc 12 resent the off-diagonal atomic blocks of the one-electron density k;m k b a matrix, and Cabc ¼ðCabcjCabcÞ, where superscripts b and a denotes subspaces of bonding and anti-bonding orbitals, respectively. The Here, the new 3cBO-scheme involves the LC coefficients given by the following direct projections: resulting three-center bonding orbitals (3) are then projected onto DE DE c the set of appropriatelyE E orthogonalized [34,35] two-center bonding ab ¼ ~fb nb ; b ¼ ~fb nb ; Tabc ab;i abc;m Tabc bc;j abc;m ~b ~b i;m j;m orbitals, fab and fbc , determined by solving the corresponding DE eigenproblems of the classical 2 2 Jug’s matrices [18] (in the same ac ¼ ~fb nb : ð Þ Tabc ac;k abc;m 13 way as in the preceding equations), i.e. k;m ! ! E E 0Da;b 0Db;c with the normalization condition ! ~fb ; ! ~fb : ð Þ y ab y bc 5 X X X Da;b 0 Db;c 0 ab 2 bc 2 ac 2 Tabc þ Tabc ¼ 1 Tabc : ð14Þ i;m j;m k;m Therefore, each 3cBO can be regarded as a linear combination of i j k orthogonal 2cBOs, It follows directly from (14) that the population of electrons origi- E E E X X fb ¼ ~fb Tab þ ~fb Tbc ; ð Þ nally assigned to the Xa–Xb and delocalized through the atomic tri- abc;m ab;i abc bc;j abc 6 i;m j;m plet Xa–Xb–Xc, calculated using the modified 3cBO-scheme (10)– i j (13), is usually lower than the corresponding electron population with the following normalization condition: from Eq. (9): X X ab 2 bc 2 X 2 2 d ab d0 d Tabc þ Tabc ¼ 1; ð7Þ ¼ k 6 N : ð Þ ; ; Nabjabc Tabc abc abjabc 15 i m j m i;m m;m i j i;m D.W. Szczepanik / Computational and Theoretical Chemistry 1100 (2017) 13–17 15
d0 Again, the diagonal matrix kabc collects the orbital populations of projecting schemes seems to increase monotonically as the size multicenter delocalized electrons assigned to each bonding orbital of molecular ring is increased. Furthermore, for the majority of spe- (10) and calculated using the procedure of canceling of non- cies, total populations of delocalized electrons calculated using the bonding and mutually phase-reversed 3cBOs, introduced in the original 3cBO-scheme (yellow bars) have substantially greater val- recent paper [5]. ues than the corresponding populations from the improved EDDB analysis (blue bars). A more detailed analysis reveals that for all the molecular systems containing one p-bond as well as for species þ 2þ 4þ Nd 3. Results and conclusions C7H7 ,C8H8 and C8H8 , the p based on the originally proposed 3cBO-scheme significantly exceeds the total number of p- 6þ 2þ To demonstrate how the reformulated 3cBO orbital-projecting electrons (red bars)! Admittedly, in the cases of C8H8 and C8H8 scheme improves the results of the EDDB analysis, let us consider the corresponding populations obtained from the improved EDDB a test set of 17 model cyclic hydrocarbons containing from three to analysis also exceed the respective formal numbers, but only by eight carbon atoms and up to three p-bonds. The EDDB density no more than 0:1%, which is actually on the very edge of the matrices (2) for each molecular system were constructed by con- assumed accuracy limit (see [5]). On the other hand, comparison sidering all possible conjugations of chemical bonds between car- of the EDDB populations from both 3cBO projection schemes with bon atoms (including the weak cross-ring carbon–carbon the corresponding total Wiberg-type bond orders [38–41] (green interactions). The values of p-component of Nd were obtained lines), calculated at the same theory level, indicates that the over- by solving the corresponding eigenproblem for Dd [2] at the estimations may come rather from the weakness of the one- B3LYP/6-31G⁄ theory level using two computational-chemistry determinant definition of the quadratic bond-order itself, espe- cially with regard to the cross-ring interactions involving carbon tools, GAMESS v5.1.2013 [36] and NBO v6.0 [37], as well as several p p scripts developed by the author. The calculated EDDB-populations atoms in meta positions [4]. Indeed, for aromatic 2 - and 6 - from both 3cBO-schemes are depicted in Fig. 1. systems the total bond-orders are generally in line with the corre- At first glance, in almost all Hückel’s aromatic systems the sponding EDDB-populations from the original formulation. The effectiveness of p-electron delocalization predicted by both 3cBO only exception is the benzene molecule, for which the magnitude of the meta-C C interaction is well-known to be very small [6,42]. This observation allows one to draw the conclusion that, in accordance with (15), both 3cBO-schemes provide comparable results only if the interaction between terminal atoms in each pos- sible pair of adjacent conjugated bonds can be neglected, which is not the case for highly ionized rings, regardless of their size [4]. Interestingly, the use of the improved EDDB scheme shows clearly that, for the Hückel’s aromatic systems, the effectiveness of bond conjugation increases logarithmically as the size of molecular ring is increased, while in the case of Hückel’s antiaromatics it displays
an exponential relationship, but only after excluding the C4H4 molecule. Due to symmetry of both occupied molecular p- orbitals, in the case of cyclobutadiene both EDDB formulations properly predict the lack of any delocalized bonding. As a matter of fact, this archetypical antiaromatic molecule at its equilibrium geometry does not exhibit neither p- nor r-delocalization and its extreme destabilization energy has been proved to be caused by a combination of angle strain, torsional strain, and Pauli repulsion [43–45]. It has to be stressed once again that for benzenoid-like systems containing three molecular orbitals of p-type the resulting effec- tiveness of p-delocalization is nearly the same regardless of the version of the 3cBO-scheme used in the EDDB definition. Indeed, for the test set of 26 six-membered polycyclic aromatic hydrocar- bons (PAH), listed in Table 1 and containing 82 unique benzenoid
Table 1 The list of 26 polycyclic aromatic hydrocarbons.
No Molecule No Molecule 1 Benzene 14 1,2–3,4–5,6–7,8-Tetrabenzoanthracene 2 Naphthalene 15 1,2–4,5-Dibenzopyrene 3 Anthracene 16 1,2-Benzopyrene 4 Tetracene 17 1,2-Tetraphene 5 Pentacene 18 1,2,3,4-Dibenzanthracene 6 Hexacene 19 1,2–6,7-Dibenzopyrene 7 Heptacene 20 1,2,5,6-Dibenzanthracene Fig. 1. Total populations of p-electrons delocalized through a system of conjugated d 8 Phenanthrene 21 3,4-Benzopyrene bonds, Np, obtained from both the original EDDB method (yellow bars) and the 9 Pyrene 22 1,2,7,8-Dibenzanthracene improved one (blue bars) for model monocyclic hydrocarbons and their ions. For 10 Chrysene 23 1,2-Benzotetracene comparison, the sums of the Wiberg-type bond orders has been added (green solid ⁄ 11 Triphenylene 24 Picene lines). Method: B3LYP/6-31G , equilibrium geometries. (For interpretation of the 12 Pentaphene 25 2,3,7,8-Dibenzophenanthrene references to color in this figure legend, the reader is referred to the web version of 13 Coronene 26 2,3,5,6-Dibenzophenanthrene this article.) 16 D.W. Szczepanik / Computational and Theoretical Chemistry 1100 (2017) 13–17 units, the improvement of the 3cBO-scheme has no noticeable effect on the results of the EDDB analysis (B3LYP/6-31G⁄), as pre- sented in Fig. 3a. That is to say, due to relatively weak cross-ring interactions between carbon atoms in meta positions, total popula- tions of electrons delocalized in each benzenoid-like unit of partic- ular PAH calculated using the modified EDDB formalism take only slightly greater values than those obtained using the original one, but the correlation between them is still very tight (R2 ¼ 0:99). In contrast, for non-benzenoid cyclic systems the EDDB-populations based on the original 3cBO-scheme tend to be somewhat overesti- mated, regardless of the number of p bonds in a molecular ring. Consequently, one should avoid comparing the EDDB-derived total populations for aromatic rings of different size when the previous 3cBO projection procedure is used. For demonstration purposes, let us consider the results of the EDDB analysis (at the CAM-B3LYP/ def2-TZVPP theory level) for the test set of 25 monocyclic species from Fig. 2, which includes mainly five- and seven-membered molecular rings (this test set comes from the original work [46]). Although the correlation between total populations of delocalized electrons from both 3cBO-schemes is maybe not the worst (R2 0:81), as shown in Fig. 3b, the use of the original formulation of the EDDB clearly fails to correctly describe the effectiveness of bond conjugation simultaneously in five- (points over the trend- line) and seven-membered rings (points under the trendline).
Fig. 3. Correlation analyses between total EDDB-populations calculated using the original 3cBO ! 2cBO projection scheme and the newly proposed one, for (a) the 4. Summary test set of 82 symmetry-unique benzenoids from molecules listed in Table 1 and (b) the test set of 25 monocyclic species presented in Fig. 2. Methods: (a) B3LYP/6-31G⁄ In this paper a new development of the electron density of delo- and (b) CAM-B3LYP/def2-TZVPP, equilibrium geometries. calized bonds formalism has been proposed to remedy some seri- ous deficiencies emerging within its original formulation [2] and It has to be mentioned that very recently an alternative scenario regarding the description of electron delocalization in aromatic of removing the electron overcounting in the EDDB method has rings. It has been demonstrated that taking explicitly into account been proposed in which additional projection onto the subspace all the trough-space interactions within the definition of three- of occupied molecular orbitals is used to eradicate non- center bond orbitals gives rise to systematic improvement of the orthogonalities in the EDDB density matrix from (2) [47]. For most results of the EDDB analysis for aromatic rings in which the of organic aromatics this particular technique gives very similar cross-ring interactions tend to play a vital role. In particular, the results as presented in this work. However, in some difficult cases EDDB-populations from the reformulated formalism do not exceed of multicenter bonds (e.g. small all-metal clusters in triplet states) the total number of p-electrons in a molecular system, which was it systematically underestimates the populations of electrons delo- sometimes observed for the corresponding populations from the calized between adjacent bonds. Since this ‘undercounting’ is prob- previous formulation of the theory. ably caused with neglecting the through-space interactions between terminal atoms in the definition (4), we are planning to combine both scenarios in the next implementation of the EDDB method.
Acknowledgments
The author is very grateful to Prof. Miquel Solà, Prof. Tadeusz M. Krygowski, and Prof. Halina Szatyłowicz for support and multiple stimulating discussions. This research was supported in part by the Foundation for Polish Science (FNP START 2015, stipend 103.2015), National Science Centre, Poland (NCN SONATA, grant 2015/17/D/ST4/00558), as well as the PL-Grid Infrastructure, with the calculations performed on Zeus: HP Cluster Platform of the Academic Computer Centre CYFRONET.
References
[1] D.W. Szczepanik, A new perspective on quantifying electron localization and delocalization in molecular systems, Comput. Theor. Chem. 1080 (2016) 33– 37. _ [2] D.W. Szczepanik, M. Andrzejak, K. Dyduch, E. Zak, M. Makowski, G. Mazur, J. Mrozek, A uniform approach to the description of multicenter bonding, Phys. Fig. 2. The test set of 25 selected aromatic and antiaromatic species. Chem. Chem. Phys. 16 (2014) 20514–20523. D.W. Szczepanik / Computational and Theoretical Chemistry 1100 (2017) 13–17 17
[3] M. Andrzejak, D.W. Szczepanik, Ł. Orzeł, The lowest triplet states of bridged [26] R. Heinzmann, R. Ahlrichs, Population analysis based on occupation numbers cis-2,20-bithiophenes – theory vs experiment, Phys. Chem. Chem. Phys. 17 of modified atomic orbitals (MAOs), Theor. Chem. Acc. 42 (1976) 33–35. (2015) 5328–5337. [27] I. Mayer, Atomic orbitals from molecular wave functions: the effective [4] D.W. Szczepanik, Evaluation of Aromaticity using the Electron Density of minimal basis, J. Phys. Chem. 100 (1996) 6249–6257. Delocalized Bonds Method.