Institute of Theoretical Chemistry Director: Prof. Dr. Christel M. Marian

SCIENTIFIC REPORT 2002 1 Staff 1.1 Prof. Dr. Christel M. Marian cm@theochem.uni-duesseldorf.de

Dr. Timo Fleig timo@theochem.uni-duesseldorf.de

Dipl.-Chem. Andreas Heßelmann andreas@theochem.uni-duesseldorf.de

Dipl.-Chem. Martin Kleinschmidt mk@theochem.uni-duesseldorf.de

Dipl.-Chem. Stephan Raub raub@theochem.uni-duesseldorf.de

Dipl.-Chem. Frank Schneider (until 08/31) institut@theochem.uni-duesseldorf.de

Dr. Thomas Schönherr ts@theochem.uni-duesseldorf.de

Dipl.-Chem. Jörg Tatchen joerg@theochem.uni-duesseldorf.de

Dipl.-Chem. Martin Torheyden martin@theochem.uni-duesseldorf.de External Collaborators Fraunhofer Institute Algorithms and Scientific Computation (SCAI)

Dipl.-Chem. Markus Doerr doerkinus@theochem.uni-duesseldorf.de

Dr. Marcus Gastreich (until 06/30) institut@theochem.uni-duesseldorf.de

Dr. Silke Reinhardt silke@theochem.uni-duesseldorf.de

1.2 Prof. em. Dr. Hans-Herbert Schmidtke schmidtke@theochem.uni-duesseldorf.de

2 Research projects 2.1 Relativistic quantum chemistry 2.1.1 Development of methods and programs 2.1.1.1 Spin-orbit coupling kit (SPOCK) The inclusion of spin orbit coupling effects in quantum chemical calculations has been a subject of intense research in our group for several years. Current work focuses on extending the applicability of the established methods to larger molecules. One major problem is the treatment of electron correlation which is tackled by the DFT/MRCI program (developed by S. Grimme and M. Waletzke) in an elegant way: dynamic correlation is accounted for by using density functionals, and a short CI expansion deals with static correlation effects. The SPOCK (Spin Orbit Coupling Kit) package, currently under development in our group, extends the DFT/MRCI package to the calculation of spin-dependent operators. Matrix elements over CI wavefunctions are available. A spin-orbit CI, treating electron correlation and spin-orbit coupling simultaneously, is under development. Central areas of application are a) compounds including heavy elements where spin-orbit coupling effects and electronic

excitations are on the same energy scale as electronic excitations, and b) calculation of spin-dependent properties without explicit sum over states. Another problem, we are addressing, is the selection of important configurations in the CI step which is conventionally done using only spin-free energy criteria. Now, single excitations that only contribute to spin-orbit coupling may be included in the calculation. An interface for the DFT/MRCI program to the MOLCAS Program has been written. Advantages of the MOLCAS interface are the inclusion of scalar relativistic effects and the availability of CASSCF wavefunctions as basis for the CI expansion. Reference: [1] S. Grimme, M. Waletzke, A combination of Kohni Sham density functional theory and multi-reference configuration interaction methods, Journal of Chemical Physics 111 (1999) 5645. Keywords: Spin-orbit coupling, spin-orbit CI, electron correlation, spectroscopy Contact: Martin Kleinschmidt, Jörg Tatchen, Christel M. Marian 2.1.1.2 Calculating ISC rates Inter-system crossings (ISC) are radiationless transitions from one electronic state to another one of different multiplicity. At closer look, also the nuclear part of the wavefunction has to be taken into account. Then ISC means a transition from one vibronic level of the initial state to isoenergetic vibronic levels of the final electronic state which often (in the case, of a large

energy gap and many degrees of freedom) form a quasi-continuum. In this case the rate of ISC can be calculated by Fermi's Golden Rule. To obtain ISC rates from theory in the Golden Rule sense, the vibronic level density and the Franck-Condon integrals entering the average squared coupling matrix element have to be known. Our program is based on the harmonic description of potential energy surfaces accessible from (standard) electronic structure program packages. Franck-Condon integrals are calculated efficiently in the displaced oscillator approximation. Vibronic levels within a chosen energy interval are obtained by a fast scanning algorithm. The program is currently interfaced to the TurboMole package and the Gaussian program. Keywords: Intersystem-crossing, density of vibrational states, Franck-Condon

integral, non- radiative transitions Contact: Jörg Tatchen, Christel M. Marian 2.1.1.3 Relativistic MCSCF program The central goal of this research project is the establishment of a program system for the computation of electric, magnetic, and spectroscopic properties of heavy elements and their molecular compounds. The implementation is intended to allow for an extensive treatment of electron correlation in a 4-component (and also approximate 2-component) relativistic framework and to make systematic approximations to both the relativistic approach and the correlation treatment possible on the same footing. In the course of 2002, a 4-component MCSCF program [1] has been brought to completion using the CI program GOSCI [2] in an upgraded version [3], allowing for determinant expansions of roughly up to 300.000 terms. The spinor optimization is carried out in the traditional picture of Dirac-Hartree-Fock (DHF) theory where the negative energy solutions are kept unoccupied and a minimax principle for the spinor rotations ensures convergence to the desired electronic state with maximum variational freedom. Double point group and time-reversal symmetries are exploited throughout for computational savings, and a quaternion algebra formalism ensures an efficient storage and addressing of the additional classes of two-electron integrals in the relativistic case. Several pilot applications have been carried out with the newly developed code [1]. The multi-configurational character of the Beryllium atom in its ground state is investigated using active spaces of increasing size. Spinor/orbital occupation numbers are compared to those obtained with a non-relativistic program package (Molcas [4]) and found to match these very well as expected in such a "non-relativistic'' system. Further, a finite-field study of the molecular dipole moment of the HBr and HI molecules is carried out, demonstrating the sufficiently improved results upon opening a suitable active space for valence correlation as compared to DHF and experimental values. References: [1] J. Thyssen, H.J. Aa Jensen, T. Fleig: A four-component relativistic multi-configuration self-consistent-field program for molecules: Implementation and initial applications, manuscript, 2003. [2] J. Thyssen: Development and Applications of Methods for Correlated Relativistic Calculations of Molecular Properties, Dissertation, Department of Chemistry, University of Southern Denmark, 2001.

Keywords: Relativistic MCSCF, 4-component, program development, heavy elements, configuration interaction, electric properties, electron correlation

Contact: Timo Fleig

(in collaboration with Prof. H.J. Aa Jensen, University of Southern Denmark, Odense) 2.1.1.4 Relativistic CI program using spinor basis The relativistic double group CI program LUCIAREL [1] has been extended by an excitation class formalism [2], interfaced to the DIRAC [3] program environment, and implemented as a CI module in the above--mentioned MCSCF code. Within this framework, it can be run as a standalone program, i.e. by truncating the MCSCF procedure after the integral transformation and the CI start guess.. The previously implemented excitation class formalism "borrowed'' from coupled-cluster theory [4] allows for an efficient implementation of a general relativistic Hamiltonian in second quantization by its constituent physical terms, e.g. the non-relativistic part, an effective one-particle (mean-field) spin-orbit part, spin-orbit two-particle terms etc. and the "occupation'' of the involved creation/annihilation operators in terms of general active spaces. Especially, the formalism enabled a concise implementation for evaluating projected (sigma) vectors and general n-particle density matrices with the same set of routines. The so-generalized program version has been applied to the three heavy atoms Tm, Lu, and Tl, where the ground-state spin-orbit splitting was analyzed depending on the type of one-particle basis: relativistic spinors or scalar relativistic orbitals. In truncated CI calculations using restricted active spaces, the spinor basis turns out to be superior to the orbital basis even in the f and d elements where the difference between the radial expectation values of the valence orbitals is by far less pronounced than in the p element Tl. References: [1] T. Fleig, J. Olsen, C.M. Marian: The generalized active space concept for the relativistic treatment of electron correlation. I. Kramers-restricted two-component configuration interaction, J. Chem. Phys. 114 (2001) 4775. [2] T. Fleig, J. Olsen, L. Visscher: The generalized active space concept for the relativistic treatment of electron correlation. II: Large-scale configuration interaction implementation based on relativistic 2- and 4-spinors and its application, in preparation, J. Chem. Phys., 2003. [3] T. Saue, V. Bakken, T. Enevoldsen, T. Helgaker, H. J. Aa. Jensen, J. Laerdahl, K. Ruud, J. Thyssen, L. Visscher: "dirac, a relativistic ab initio electronic structure program", release 3.2, 2000. [4] J. Olsen: The initial implementation and applications of a general active space coupled cluster method, J. Chem. Phys. 113 (2000) 7140. Keywords: Relativistic CI, spinor basis functions, large-scale configuration interaction,

excitation operators, active spaces, sigma vectors, density matrices, spin-orbit coupling

Contact: Timo Fleig

(in collaboration with Prof. J Olsen, University of Aarhus, Denmark and Dr. L Visscher, Free University of Amsterdam, The Netherlands)

2.1.1.5 Development of a driver for the numerical determination of minima on MRCI potential energy surfaces

Nowadays, equilibrium geometries are routinely determined by analytical gradients only for electronic ground states and the lowest state of a given spin and space symmetry. Methods that can be employed also for electronically excited states mostly suffer from an insufficient treatment of either dynamic correlation (e.g. CASSCF) or static correlation (e.g. TDDFT) or both (e.g. CIS). We have developed a program that determines numerical gradients from finite differences on a grid of Cartesian coordinates and searches for minima on potential energy surfaces. A parallelized version of the program utilizes the MPI protocol. Currently available search algorithms are steepest descent and conjugate gradient. In principle, the driver can be coupled to any quantum chemical package. Presently, drivers exist for Turbomole and the DFT/MRCI program. Employing this tool, we have optimized the geometries of several excited singlet and triplet states of uracil and adenine. Keywords: Numerical gradient, geometries, electronically excited states Contact: Frank Schneider, Christel M. Marian 2.1.2 Spindependent phenomena in organic molecules The photochemical and photophysical activity of organic molecules decisively depends on the effectivity of spinforbidden radiative and nonradiative transitions, i.e., phosphorescence and intersystem crossings (ISC). After photoexcitation, the comparably long-lived lowest-lying triplet state (T1) often acts as an energy sink from which photochemical reactions proceed. In our group, efficient methods are developed for the quantum chemical description of spinforbidden processes in larger molecules. First calculations on thiocarbonyl compounds show promising results. Applications focus on psoralens that are of high pharmacological interest (photodynamic therapy) and the nucleo bases uracil, thymine, and cytosine. A further focal point is the computation of g-tensor shifts in comparison with ESR spectroscopy. Keywords: Spin-orbit coupling, phosphorescence, intersystem crossing, spectroscopy,

photochemistry,photophysics, g tensor, configuration interaction, electron correlation, quantum chemistry

2.1.2.1 Electronic spectrum and phosphorescence of dithiosuccinimide The dithiosuccinimide molecule exhibiting intense phosphorescence forms a realistic example to test the predictive power and performance of the SPOCK spin-orbit coupling kit together with DFT/MRCI. Due to the presence of two coupled C = S groups, the electronic spectrum contains two near-degenerate triplet n →π* states. Therefore, it is difficult to calculate the

adiabatic order of states with high accuracy and decide which is the phosphorescing one. To this end, we performed TD-DFT geometry optimizations for the low-lying electronic states in question.

Fig 1. The dithiosuccinimide molecule Both triplet n →π* states show bond elongation in the C = S groups, but do not deviate noticeably from C2v symmetry. Our calculated radiative life-time of the lower 13B1 n →π* state amounts to 0.5 ms in excellent agreement with the experimental value. The higher 13A2 n →π* state has an even larger spin-orbit coupled dipole transition moment to the singlet ground state. This finding explains certain previously unresolved peculiarities in the experimental phosphorescence excitation spectrum. Keywords: Spin-orbit coupling, phosphorescence, thiocarbonyles Contact: Jörg Tatchen, Martin Kleinschmidt, Christel M. Marian (in cooperation with Maja Parac and Stefan Grimme, University of Münster) 2.1.2.2 Spin-forbidden transitions in thio- and seleno-psoralens Naturally occurring psoralens are used as photobiologically active drugs against skin diseases (psoriasis, vitiligo). Currently, considerable effort is spent to extend the pharmaceutical potential of psoralens towards the treatment of skin cancer and several virus diseases. Synthetic psoralens in which one or both intracyclic oxygen atoms are substituted by sulphur or selenium (Fig. 2) offer a key to optimizing the therapeutical activity. It is commonly believed that photoexcited psoralens react mainly from the triplet T1 state. The efficiency of inter-system crossing (ISC) from a photoexcited singlet state into the triplet manifold and the life-time of the T1 state should thus be of crucial importance. The theoretical exploration of hetero-substituted psoralens is the concern of this project.

Fig 2. Naturally occurring psoralens: X,Y=O. In synthetic thio- and seleno-psoralens X, Y, or both =S,Se.

Fig 3a. Vertical excitation energy for the lowest singlet and triplet states of heteropsoralens

(DFT/MRCI, TZVP basis set at the KS-DFT (BLYP) optimized ground state geometry).

Fig 3b. Absolute value of spin-orbit matrix-elements (SPOCK package, xy: sum of HSO,x and

HSO,y. Abscissa coding: hetero atoms XY according to Fig. 2. Keywords: Spin-orbit interactions, intersystem crossing, photochemistry, photodynamic

therapy, psoralenes Contact: Jörg Tatchen, Christel M. Marian

2.1.2.3 Electronic excitation and singlet-triplet coupling in uracil tautomers and uracil-water complexes

Electronic spectra of uracil in its diketo (lactam) form and five enol (lactim) tautomeric forms have been investigated by means of combined density functional and configuration interaction methods. Local equilibrium geometries of the electronic ground state and several low-lying excited states have been optimized. In accord with experiment, we find the diketo tautomer to be the most stable one. The calculations confirm that the first absorption band arises from the 1(π → π*) S0 → S2 excitation. The experimentally observed vibrational structure in this band originates from a breathing mode of the six ring. Among the uracil lactim tautomers, one is particularly interesting from a spectroscopic point of view. In this tautomer, the π → π* excitation gives rise to the S1 state. We have simulated the effects of hydrogen bonding with a protic solvent by recomputing the spectrum of uracil in the presence of two, four, or six water molecules. Complexation with water molecules is seen to cause a significant blue shift of n → π* excitations while leaving π → π* excitations nearly uninfluenced. Spin-orbit coupling was computed by means of the SPOCK code (cp. 2.1.1.1) for DFT/MRCI wavefunctions employing a non-empirical spin-orbit mean-field approach. Radiative life-times have been determined by perturbation theory for the experimentally known weak phosphorescence from the π → π* excited T1 state. For more information see paper [6.1.7]. Keywords: RNA/DNA bases, fluorescence, phosphorescence, radiationless transitions,

quenching Contact: Christel M. Marian, Frank Schneider, Martin Kleinschmidt, Jörg Tatchen 2.1.3 Very precise calculations on small molecules 2.1.3.1 Ab initio study of the vibronic and spin-orbit structure in the X2Π

electronic state of CCCH Potential energy surfaces and coupling matrix elements for the electronic states of the CCCH radical, correlating at linear nuclear arrangement with the X2Π state, have been calculated by means of an extensive ab initio CASSCF/MRCI approach. They are used to compute the vibronic and spin-orbit structure of this electronic state by means of a variational approach based on the use of normal bending coordinates [6.2.3]. The results of calculations enable a reliable interpretation of the available experimental data and offer an explanation for several apparently peculiar features observed. In cases of relatively small values for the Renner parameters, the spin-orbit splitting in so-called unique levels is generally comparable to the spin-orbit coupling constant of the linear molecule. In Π electronic states of triatomic molecules these are the lowest-lying K ≠ 0 vibronic levels. In tetra-atomics there are several levels of unique character for each value of K ≠ 0. In nonunique K ≠ 0 vibronic levels, the spin-orbit splitting is more or less efficiently quenched. However, in the present case, characterized by an extremely flat CCC-H bending

potential curve for the A' component of the X2Π state and large values of Renner parameters, the classification of vibronic levels in unique and nonunique ones is not unambiguous. This also concerns the lowest-lying vibronic level, K = 1(v4 = 0, v5 = 0). The ab initio computed spin orbit splitting of this level is 13.74 cm-1 , i.e., only about half the value of the spin-orbit constant. This result is in excellent agreement with the corresponding experimental finding of 14.4 cm-1 and explains thus the apparent disagreement between previous theoretical and experimental results. Keywords: Renner-Teller coupling, finestructure splitting, vibronic interaction Contact: Christel M. Marian

(in collaboration with Miljenko Perić, Mirjana Mladenović, and Katarina Tomić, University of Belgrade, Yugoslavia)

2.1.3.2 Radiative transitions in the ICN molecule The photodissociation of the ICN molecule is one of the first problems that have been investigated by femtosecond laser techniques by Zewail and coworkers in the 1980s. Among other reasons, this choice was motivated by the fact that this triatomic molecule could readuly allow a comparison with theory. However, the large spin-orbit and correlation effects in this system are not an easy task for theory. There is some controversy about the transition dipole moments for the lowest excited states of Π type. We have performed an all electron calculation of the spin-free spectrum of ICN at the DFT/MRCI level. The non-empirical one-center mean-field spin-orbit Hamiltonian (Breit-Pauli) has been employed. The spin-orbit coupled spectrum has been obtained at the level of quasi-degenerate perturbation theory in the basis of LS-coupled states. According to our calculations, the most intense transition in the energy region below 6eV stems from the 1Π state owing a transition dipole moment of 0.33 a.u. to the ground state X 1Σ+. The X 1Σ+ → 1Π0+ transition which is mentioned as the stronger one in the literature comes out with much less intensity (0.08 a.u.). Keywords: Spin-orbit coupling, excited states, electronic spectrum Contact: Jörg Tatchen, Christel M. Marian 2.1.3.3 Dipole moments in excited states of ScO The 2Π excited state components of the ScO molecule exhibit a significant difference of 0.37 D in their experimental dipole moments [1]. In these light elements, where relativistic effects are small, such a difference can only be explained by the coupling of the (3/2) component of a very close-lying 2∆ state to the (3/2) component of the 2Π state. The description of these states including their dipole moment by a finite electric field approach is currently being carried out with the CI programs LUCITA [2] and LUCIAREL [3] using a common set of orbitals for all involved states and by including spin-orbit coupling in a variational procedure on the same footing with electron correlation. The computations are costly (up to 30 million determinants in the spin-dependent case), require a careful construction of the active excitation spaces, and

comprise model calculations for general CI applications of this type to difficult open-shell systems where relativistic effects are important. References: [1] J. Shirley, C. Scurlock, T. Steimle: Molecular-beam optical Stark spectroscopy of ScO, J. Chem. Phys. 93 (1990) 1568. [2] LUCITA is a direct CI program written by J Olsen, Molcas interface by T. Fleig, 1999, dirac interface by T. Fleig, 2001. [3] LUCIAREL is a direct, relativistic double group CI program written by T. Fleig and J. Olsen, Molcas inter-face by T. Fleig, 2000. Keywords: Dipole moment, excited states, spin-orbit coupling, large-scale configuration

interaction, open-shell systems Contact: Timo Fleig

(in collaboration with V. Kelloe and. M. Urban, University of Bratislava, Slovakia) 2.2 Properties of large molecules and solid state materials 2.2.1 Modeling high-performance ceramics High-performance ceramics that contain the elements Si, N, B, and C are employed as high-temperature stable fibers in composit materials. For a better understanding of their physical properties and to enhance their performance features, it is necessary to understand the atomistic structure of these materials. Since the ceramics possess no translational symmetry, the structure determination by means of experimental methods only is difficult. Computer simulations can give clues for the solvation of this problem. To this end, we pursue complementary approaches. 2.2.1.1 Investigation of reactions of molecular precursors One way to get information about the atomistic structure and preferred coordinations in the amorphous network, is a quantum chemical investigation of the reactions during the cross-linking of the precursor. If it is known which processes are preferred, it can be concluded which atomistic patterns can be found in the final ceramic.

The molecule TSDE For a survey over the different reactions of the molecular precursor 1-(trichlorosilyl)-1-(dichloroboryl)ethane (TSDE) with methylamine, thermodynamical data of the gas-phase reactions (heats of reactions and barrier heights) have been calculated by solving an approximated, time-independent Schrödinger equation (RI-MP2). Reactions proceed along specified courses in this cases. In experiment, the linkages of the molecular precursors are conducted in solution and not in gas phase. For a proper investigation of solvent effects, Car-Parrinello molecular dynamics (CPMD) simulations have been performed. However, simulation time scales have prevented the observation of a spontaneous reaction between CH3NH2 and TSDE. Therefore, the influence of an excess of methylamine on the energetic course has been investigated by calculating constrained minimum energy paths. To receive an impression of the extent of entropic effects during these reactions, free energy courses have been estimated by means of the Blue Moon ensemble. A fundamental observation during the investigation was that solvent molecules participate actively in the reaction and that the mechanisms change. Furthermore it was found that some processes are favored more by solvent effects than others. The calculations have shown that the Si-Cl/B-Cl bonds break during the first steps of the polymerisation. The internal Si-C/B-C bonds of the precursor are not destroyed. Reactions at the boron center of the precursor are preferred over reactions at the silicon side. Therefore, the precursor units are linked more frequently via their boron sides. Such a preferred reaction pathway can lead to an to a nonuniform distribution of boron and silicon in the final ceramic. Keywords: Car-Parrinello molecular dynamics, solvent effects, ceramics, TSDE, entropy Contact: Silke Reinhardt, Christel M. Marian 2.2.1.2 Development of structural models for amorphous solids using

Reverse Monte Carlo Methods

The Reverse Monte Carlo Method is a widely used approach for modeling the structure of amorphous solids. Basic idea of this method is to modify a structural model randomly until the agreement between data calculated from the model and experimental data is as good as possible. Usually, the experimental data is taken from diffraction experiments and represents only onedimensional projections of the threedimensional structure. Besides that, experimental data is prone to artefacts resulting from the transformation between the reciprocal space (scattering intensities) and the real space (radial distribution functions). To improve the reliability of the structural model, it is reasonable to take data from various experiments into account and to reproduce the process of experimental data acquisition as exactly as possible. Another problem arising in developing of structural models using the Reverse Monte Carlo Method is the occurrence of unrealistic, energetically unfavorable structures. This problem may be solved by introducing additional constraints such as interatomic minimum distances or coordination number distributions. However, a more elegant way is the direct evaluation of the potential energy during the Reverse Monte Carlo optimization. We have developed a Reverse Monte Carlo program that allows the usage of NMR data in addition to diffraction data from X-ray, electron, and neutron scattering experiments. The diffraction data may be calculated either in the real space or in the reciprocal space followed by a Fourier transformation. In each optimization step, the potential energy of the model may be calculated. Keywords: RMC, Reverse Monte Carlo, amorphous solids Contact: Markus Doerr, Christel M. Marian 2.2.1.2 Parameterization of Coulombic potentials for ternary high-

performance ceramics A prerequisite for the structural modeling of amorphous solids are energy functions which may be evaluated rapidly - this is especially true for large-scale simulations. Quantum che-mical approaches are prohibitive from several hundred particles upwards, so one often makes the compromise to utilize so-called model potentials, the latter being parameterized on the grounds of both experimental and/or calculated data. In the framework of the DFG center of excellence SFB 408 ("Inorganic Solids Lacking Translational Symmetry"), we have parameterized such potentials for the ternary system Si/B/N. During the period which this report covers, this is especially true for finishing an efficient Coulombic potential (termed "Qi"); treating atomic charges explicitly complements the existing set of potentials. During the evaluation phase, it has become obvious that the quality of predictions for solid state properties is significantly improved: Besides structural data which is ameliorated, the so-called second-order properties are improved, i.e., bulk moduli, vibrational frequency and other curvature-dependent features result with higher accuracy than before. The analytic expressions for two-body terms are of the Buckingham form, three-body contributions have been expressed as Stillinger-Weber terms. The standard problem of artificial atom collapse (associated with short distances when using Buckingham-type potentials) has been solved by introducing a damped dispersion expression according to Tang and Toennies. Similarly, in the longer range region, the twobody expressions have been tapered such that no singularities on the hypersurface are created (otherwise, the potential would only have limited applicability in simulations, esp. MD simulations.)

To our knowledge, this potential is the first non-quantal potential which correctly mimics the energetic bilance of the binary (B/N, Si3N4) w.r.t. a hypothetical ternary (Si/B/N) phase.

Run of the three-body part in the potential Qi.

The damping with interatomic distance r12 becomes obvious.

References: [6.3.2] Keywords: amorphous ceramics, model potentials, structural modeling Contact: Marcus Gastreich, Christel Marian 2.2.1.4 QSPR for Solids Based on Nitrogen-NMR Properties A QSPR method (quantitative structure property relationship) we had developed for boron nitrides [1] has been partially transferred to the much more complicated structures of silicon nitrides. Outcomes of QSPR investigations may perhaps serve as an additional information in cost functions for the modeling of amorphous solids. The successful validation phase of the project has been based on smaller Si/N-compounds, as for example found in Bettina Jaschke's PhD theses [1]. The computational procedure was as follows: For the most frequent polymorphs of silicon nitride (alpha- and beta-), we cut out "clusters" from the solid which are centered around one of the crystallographic different nitrogen atoms. For the central atoms, nitrogen NMR shifts were determined. All clusters have been H-saturated, optimized, and checked for the convergence of chemical shifts w.r.t. their size. In fact, for the first time, quantum chemical data could be attributed unambiguously to the two crystallographically different nitrogen atoms in beta-Si3N4. Since the solid state structures of silicon nitrides are by far more complicated than the ones of boron nitrides, the coupling of the degrees of freedom when distorting the cut-outs is much stronger. This is why the analyses [3] of variations of NMR-shifts with structure, i.e., the QSPR itself, is work waiting to be done in more detail in the near future.

Cut-out for the determination of N-chemical shifts for beta-silicon nitride. The "atom-in-focus" (and its binding directions), is tinted yellow.

N and Si alternate in the solid state structure. References: [1] see [6.3.2]. [2] M. Gastreich, C.M. Marian, Physical Chemistry, Chemical Physics 2 (2000) 955. [3] Bettina Jaschke, Dissertation University of Göttingen, 1999. Keywords: QSPR, NMR, QSPR, boron nitride, silicon nitride Contact: Markus Doerr, Marcus Gastreich, Christel Marian 2.2.2 Transition metal compounds with specific optical and magnetic

properties 2.2.2.1 Cyano complexes of chromium(III) The goal of this work was to get a better understanding of the M-CN bonding in monomeric cyano complexes in order to help to elucidate the electronic and geometrical properties in binary cyano complexes, and CN-bridges containing clusters and networks as well. For our investigation we have selected the octahedrally coordinated anion [Cr(CN)6]3- together with the tetragonal cations [Cr(NH3)5CN]2+ and trans-[Cr(NH3)4(CN)2]+, because detailed spectroscopic material is available in these cases [1]. By experimental and theoretical analyses was shown how the Cr-CN bonds are influenced by the other ligators of the chromophore. For example, according to a DFT analysis, the CN ligand in [Cr(CN)6]3- behaves as a moderate π-acceptor, i.e. eπCN

< 0 in the framework of the Angular Overlap Model (AOM [2]). This is in accordance with results obtained from a ligand field treatment. On the other hand, if "hard" ligands like NH3 replace one or more cyano ligands, an increase of the π-donor contributions of the TM-CN bond is observed. In fact, this may lead to a weak net donor behavior as determined for the pentaammine cyano complex (eπCN > 0).

2.2.2.2 The photo-isomerization process in Fe(III) nitroprusside Upon irradiation with visible light the red colored compound Na2[Fe(CN)5NO] with a nitrosyl Fe-NO bond in the electronic ground state can be transformed with up to 40-50% yield into metastable excited states. At temperatures below 150 K the system can freeze out and reside in these states for a very long time, before reversibly converting back, either thermally or by excitation with red light. The color change associated with the generation of these excited states has been proposed as a basis for reversible storage and reproduction of information using light at two different wavelengths. The aim of our work was to elucidate the states involved in the photo-isomerization process. The adiabatic potential energy surface of the ground state (GS-APES) of the anion [Fe(CN)5NO]2- has been studied by means of DFT. A decomposition of the total bonding energy into orbital interaction, electrostatic bonding and exchange (Pauli) repulsion terms (extended transition state method, ETS) has been carried out and the chemical bonding at all stationary points of the ground state potential surface (minima and saddle points) has been characterized. On this basis, the unique topology of the electronic ground state with minima for linear Fe-N-O (ground state, GS) and Fe-O-N (metastable state, MS1) and side-on Fe-(NO) bonded (metastable state, MS2) has been rationalized. The potential energy surface of the lowest excited [3d→π*(NO)] electronic state (ES-APES) with minima at non-linear Fe-N-O and Fe-O-N arrangements has been calculated starting from higher symmetric (C4v) structures (linear Fe-N-O and Fe-O-N bond) utilizing a 1E Jahn-Teller coupling mechanism. We found that after geometrical relaxation, the ES-APES crosses the GS-APES in the vicinity of the transition states, thus affecting the photo-isomerization process. In the light of these results a combined ground and excited state configurational energy diagram was presented which can be used to discuss the mechanism of photo-isomerization [3]. 2.2.2.3 TM-doped KTP crystals and oxyborates TM ions doped into crystal lattices may probe the local environment. In the present case the geometric deviation of the Ti ions from the regular high-symmetry position is most important, because it is responsible for the non-linear optical (NLO) properties of KTP (KTiOPO4). We have investigated optical transitions of Cr(III) and Mn(IV) doped KTP single crystals in absorption and emission and were able to interpret the d-d spectra by means of AOM reasonably. However, further geometrical distortions, which are expected when Ti(IV) ions are substituted by TM ions, could not be manifested by the analyses of the relatively broad Volume structures. For smaller KTP-related systems like, for example, [M(O)2(H2O)4] (M=Ti,Cr) we have calculated hyperpolarisizations by using the DFT method. Hereby we could show that the non-linear optical properties are strongly influenced by the additional electrons provided by the substituting TM ions. On the other hand, the second coordination sphere may also contribute significantly to the NLO effect, as can be demonstrated for the artificial trigonal complex [TiO6(TiO(TiO)6O2]. Colorless crystals of the novel oxyborate Na2Ga2(BO3)2O consist of a three-dimensional network showing trigonal symmetry. When doped with Co(II) ions, which occupy the pseudo-tetragonal Ga(III) sites, the color of the crystals changes towards blue due to d-d transitions which are localized at the cobalt centers. Volumes splittings observed in the optical absorption spectrum are reasonably explained by means of the AOM: they are induced by a reduced local symmetry that is caused by charge compensation in the coordination sphere

(Co2+ substitutes Ga3+). In order to derive more precise structural information, additional experimental data are required. Accordingly, an optical investigation of electronic transitions in the middle and near infra-red is in progress. References: [1] Schönherr, T., Itoh, M., Urushiyama, A.: Analysis of d-d transitions in trans-[Cr(NH3)4(CN)2]+ as inferred

from polarized optical spectra and angular overlap model calculation. Bull. Chem. Soc. Jpn, Vol. 68(8) (1995) 2271-2276.

[2] Schönherr, T.: Angular Overlap Model applied to transition metal complexes and dN-ions in oxide host lattices. Topics in Current Chemistry, Vol. 191 (1997) 87-152.

Keywords: Transition metal compounds, cyanide complexes, KTP, TM-doped oxides,

optical transitions, NLO, AOM, DFT Contact: Thomas Schönherr

(in collaboration with. M. Atanasov and. P. Peshev, Bulgarian Academy of Science) 2.2.2.4 Optical investigations of alkali metal thiomanganates(II) Tetra-coordinated Mn(II) complexes providing different molecular structures were investigated using various spectroscopic procedures. Na6MnS4 contains separate pseudo-tetrahedral Mn-S complex units, K2Mn2S2 has chains of edge-shared tetrahedra, and Cs2Mn3S4 crystallizes in corresponding layers. Also doped materials, i.e. Cs2(MnxZn1-x)3S4 (0.0<x<1.0) are considered. Absorption and emission spectra of neat and doped materials have been obtained. Assignments to localized d-d transitions have been derived from Angular overlap model calculations. All compounds show intense luminescence in the red, some of them also in the yellow region. Decay measurements supply lifetimes and activation energies evaluated from Arrhenius plots. We present a model which allows to describe the emission properties of these materials. Keywords: Transition metal compounds, thiomanganates, optical spectra, luminescence,

angular overlap model Contact: Hans-Herbert Schmidtke 2.3 Modeling of intermolecular interactions 2.3.1 Quantum Chemical Calculations on CYP51 The cytochrome P450 14-alpha-sterol-demethylase (CYP51, Fig. 1) belongs to the group of oxydoreductases and is involved in steroid biosynthesis, just like many other CYPs. The active site (Fig. 2) contains a substituted porphine ring with an iron at its center (FeP). The oxygen transfer to the substrate is catalyzed by the iron center. In the course of this reaction, the oxidation level of the iron changes from II to III an back to II. CYPs are present in almost any cells. The subject of interest in this study was CYP51 from fungi.

Figure 1 Figure 2 Imidazole and its derivatives are known to inhibit CYPs by coordinating the iron stronger than the oxygen. Hence, a therapeutical application of such imidazole derivatives is the use as antimycotica. The present antimycotica are mostly way too unspecific to tell apart human and fungal CYPs, so that cells of both species will perish. Such unspecific antimycotica can only be used topically (on the skin). Some disease patterns lead to systemic fungal infections. In most cases, an insufficient immune system is the reason for this (e.g. HIV, organ transplantation, etc.). So specific antimycotica are highly desired. Only a few of such specific antimycotica are available at the time and not all of them work properly. It was our aim to provide an insight into the iron-imidazole bond with the help of quantum chemical calculations on this system (FeP-imidazole) to eventually design more specific imidazole derivatives. The enzyme CYP51 consists of 455 amino acids with 3539 non-H-atoms and contains nearly 52.000 electrons. Molecules of this size cannot be treated by standard quantum chemical methods. Therefore, we have focused on the active site - the calculations were made with a model system (Fig. 3). This model system contains a porphine (unsubstituted) with a complexed iron-II (FeP-complex) in the center. In the real enzyme the iron is connected to the backbone via a cysteine which is replaced by a methylthiolat (MeS-) in the model system. In the crystal structure, the CYP51 is complexed with 4-phenyl-imidazole which is replaced by imidazole.

Figure 3

The important questions were: a) How strong is the Fe-imidazole bond? b) Is the ligand staggered or gauche regarding the Fe-N bonds in porphine? c) What is the cost of the tilting of the ligand? The degrees of freedom in the model system were chosen accordingly. The energy of the system was determined on a grid varying the length of the Fe-imidazole bond d, the rotation θ and the tilting ϕ of the ligand. All calculations were performed with density functional theory and the resolution-of-identity-approximation (RI-DFT) with the TurboMole 5.1 software suite. For every atom a TZVP basis set was used (except for a SV basis set for the hydrogen atoms). A coordinate driver “RbArr2” was programmed that generates a configuration of the model system according to the model coordinates (d, θ, ϕ) in the formate of TurboMole. Then each configuration is sent to an own node and monitored. To determine the strength of the Fe-imidazole bond configurations with various Fe-imidazol distances d were calculated to obtain an energy profile. The resulting bond strength amounts to ∆E = 47.71 kJ/Mol. Varying the angle θ we found minima at θ = 45° and θ = 135° as well as maxima at θ = 0°, 90°, and 180°. The question whether the ligand is staggered or gauche must remain unanswered. The energy barrier between these two configurations is smaller than 4.5 kJ/Mol which can easily be overcome at normal blood temperatures. Though all CYPs are very similar, crystal structures show slight differences of 5° in the tilting angle ϕ between human and fungal CYPs. The calculations show a rapid decrease in the binding energy while increasing the tilting. There is hope that this information can be used to design more selective ligands. Keywords: CYP51, imidazole, iron-porphyrine, RI-DFT, energy profile, HEM, TurboMole Contact: Stephan Raub, Christel M. Marian

(in cooperation with Bernd Rupp and Prof. H.-D.Höltje, Theoretical Pharmacy, HHUD) 2.3.2 Treatment of intermolecular interactions by means of perturbation

theory 2.3.2.1 Modeling of an intermolecular potential energy surface for the water

dimer based with symmetry-adapted perturbation theory Liquid water shows a wealth of extraordinary physical properties, such as a maximum of density at about 4 oC and, compared to liquids with a similar molecular mass, an extremely high heat capacity and boiling point. Important methods of theoretical inquiry of such thermodynamic properties of liquids are molecular dynamics- and monte carlo simulations trying to reproduce the motion of the molecules in the liquid. In such simulations the intermolecular potential, whose gradient provides the forces between the molecules in terms of their distance and their mutual orientation, is of crucial importance.

The subject of our work is the development of an intermolecular potential energy function on the basis of demanding and accurate ab initio calculations. Because of the sizable calculation time requirements per point of the potential energy surface, a relatively small amount of points is available. In order to ensure an adequate behavior of the potential energy function, the following criteria should be met by the potential energy function:

- The number of parameters determined by fitting to the ab initio points should be as small as possible.

- The mathematical form of the potential energy function should be deduced tightly from physical models for intermolecular interactions.

In order to fulfill the second criterion, the intermolecular interaction energy should be divided into the contributions of the electrostatic, the induction and the dispersion energy. Each of these contributions is accompanied by a respective exchange counterpart as a consequence of the antisymmetry of the wavefunction of the system with respect to the Exchange of electrons between different molecules. All these energy contribution have been calculated [1] at any ab initio point of the potential energy surface by means of the symmetry adapted perturbation theory (SAPT) of Jeziorski et al. [2]. So far, the fits of the exchange energy contributions, the electrostatic energy and the energy caused by the charge transfer, which has been separated from the induction energy, have succeeded. The number of parameters required in these fits is comparably small. References [1] B. Jeziorski, R. Moszynski, K. Szalewicz, Chem. Rev. 94 (1994) 1887. [2] M. Torheyden, G. Jansen, "Interaction energies for the water dimer by supermolecular methods and symmetry-adapted perturbation theory: The role of bond functions and convergence of basis subsets.", Theor. Chem. Acc. 104 (2000) 370. Keywords: Water dimer, SAPT, potential surface, intermolecular interactions Contact: Martin Torheyden, Georg Jansen 2.3.2.2 A combined density functional perturbation theory approach for

intermolecular interactions A perturbational ansatz allows one to calculate the interaction energy between molecules as a sum of electrostatic, induction and dispersion energies together with their corresponding Pauli-repulsion terms. In the limit of large monomer distances, this sum can be reduced to the well known interaction terms between multipole moments and static and dynamic polarizabilities. In case of smaller intermolecular distances new terms occur, which arise due to the overlap between the charge distributions as well as electron exchange effects, coming from the Pauli-repulsion.. All these terms can be expressed through the static and induced charge distributions as well as the corresponding density matrices of the distinct molecules. For the calculation of these properties, one is not restricted to use traditional ab initio quantum chemistry methods, but they also can be obtained from density functional theory, which has recently been proposed by us [1]. Such a combined density functional perturbation theory method circumvents the known difficulties of todays density functional methods to describe intermolecular interactions (in particular the dispersion energy) and is on the other hand far less computational expensive than the standard ab initio perturbation theory method. An analysis

of this new approach shows, however, that the Pauli-repulsion terms are not 'potentially exact' [1]. This even holds, if the static and induced molecular charge distributions are being perfectly described by density functional theory. On the other hand, the electrostatic, induction and dispersion energies are potentially exact within this theory. It has been investigated, whether the combined density functional perturbation theory method successfully predicts intermolecular interaction energies [2,3,4]. For this, a number of small test systems has been chosen, which partly only weakly interact through dispersion effects and partly through relative strong hydrogen bonds. It has been found, that most of the common density functionals fail to describe the important correlation contributions of the different interaction terms, while our own developed asymptotic exchange correlation potential serves for that purpose. The results fairly show, that indeed the density functional perturbation theory approach may successfully be applied for the calculation of interactions between extended monomers, while conventional ab initio perturbation theory methods are not feasible in this case. References: [1] G. Jansen und A. Heßelmann, J. Phys. Chem. A 105 (2001) 11156 Keywords: Intermolecular interactions, Kohn-Sham orbitals, Brueckner orbitals,

symmetry-adapted perturbation theory, SAPT Contact: Andreas Heßelmann, Georg Jansen

3 Diploma and Dissertations 3.1 Diploma Theses Frank Schneider (May 2002): "Entwicklung eines Treibers zur numerischen Bestimmung von Minima auf CI-Potenzial-flächen und Anwendung auf angeregte Elektronenzustände des Uracil-Moleküls" (Development of a driver for the numerical determination of minima on CI potential surfaces and application to excited electronic states of the uracil molecule) 3.2 Doctoral Theses Reinhardt, Silke (December 2002) "Atomare Verknüpfungen in (Carbo)-Borosilazan-Hochleistungskeramiken: Neue Erkenntnisse aus quantenchemischen Untersuchungen von Polymerisierungsreaktionen molekularer Vorläufer mit ab initio-Verfahren und Car-Parrinello-Moleküldynamik" (Atomic links in (carbo)-borosilazane high-performance ceramics: New insights from quantum chemical investigations of polymerization reactions of molecular precursors using ab initio approaches and Car-Parrinello molecular dynamics)

4 Grants 4.1 DFG - individual grants Ab initio-basierte Wechselwirkungspotentiale für wasserstoffverbrückte Systeme (Reference: Ja 954/1-1, 1-2) Prof. Dr. Georg Jansen Entwicklung und Anwendung eines Programmpakets zur effizienten Beschreibung spin-bahn-abhängiger Eigenschaften großer Moleküle (Reference: Ma 1051/5-1) Prof. Dr. Christel M. Marian Computergestütztes Templatdesign: de novo-Entwurf und rationale Optimierung (Reference: Ma 1051/6-1, applied) Prof. Dr. Christel M. Marian Transition metal compounds with specific optical and magnetic properties (Reference: Scho 320/6-1) German-Bulgarian collaboration (436 BUL 113/104/0). Dr. Thomas Schönherr 4.2 DFG - priority program Theoretical Investigation of Magneto-Structural Correlations in Inorganic Cyano-Bridged Transition-Metal Complexes (Reference: Scho 320/7-1, submitted) within the DFG priority program 1137 “Molecular Magnetism” Dr. Thomas Schönherr 4.3 DFG - Collaborative Research Centers Ab initio-Berechnung von Reaktionen molekularer Vorläufer und NMR-chemische Verschiebungen (SFB 408 “Anorganische Festkörper ohne Translationssymmetrie”, TP C2) Prof. Dr. Christel M. Marian Parametrisierung von Kraftfeldern (SFB 408 “Anorganische Festkörper ohne Translationssymmetrie”, TP C5) Prof. Dr. Christel M. Marian 4.4 Support from HBFG Hochleistungs-PC-Cluster (Reference: Hbfg 110-471) Prof. Dr. Christel M. Marian

4.5 Society of Friends and Promoters of the HHUD A grant supports the editorial work for a special series dedicated to the late Prof. C.K. Jørgensen" within the series Structure and Bonding (Springer Publishing House) Dr. Thomas Schönherr 5 Lectures and posters 5.1 Invited lectures 5.2 Other external lectures 5.3 Posters 5.3.1 Doerr, Markus; Gastreich, Marcus und Marian, Christel M. Using NMR data in a Reverse Monte Carlo Approach for Modeling Solids Euro Winter School: Quantum Simulations of Complex Many-Body Systems: From Theory to Algorithms, Kerkrade, 25.02. - 01.03.2002

Abstract: We present an extension of the Reverse Monte Carlo approach for modeling amorphous solids which allows the use of NMR-data as an additional part of the cost function. This ansatz should improve the reliability of the resulting structural model. Reverse Monte Carlo approaches suffer from ambiguities that arise because all diffraction data are 1D-projections of the 3D-structure. Therefore, it is advisable to utilise as many different experimental data as possible. Recent experience has shown that the isotropic 15N-NMR-shifts for B/N-systems may be parameter-ised as a function of the first two coordination spheres around an N [1]. These correlation func-tions have been implemented in an RMC routine. To test the feasibility of our approach, we have applied it to clusters which had been generated by cutting out atoms from both hexagonal and cubic BN. [1] C.M. Marian, M. Gastreich, Sol. State NMR 19 (2001) 1.

5.3.2 Doerr, Markus; Gastreich, Marcus; Marian, Christel M. Using NMR data in a Reverse Monte Carlo Approach for the Modeling of Amorphous Solids Methods in Molecular Simulation, CCP5 Summer School 2002, London, 08.07. - 16.07.2002

Abstract: Ab extension of the usual Reverse Monte Carlo approach is presented. Our code allows the use of NMR data and the potential energy as additional parts of the cost function. We have enhanced our code so that the scattering data may be computed either in reciprocal space from the scattering intensity or in real space from the radial density. To test the applicability of our approach, we have applied it to clusters cut out from hexagonal and cubic BN.

5.3.3 Heßelmann, Andreas; Jansen, Georg Ein kombiniertes Dichtefunktional-Störungstheorieverfahren für intermolekulare Wechsel-wirkungen Symposium "Molekulare Erkennung", Essen, 07.-09. 03. 2002

Abstract:

Die meisten der in der Chemie gängigen Dichtefunktionalverfahren sind darauf optimiert, die Elektronenstruktur im Valenzbereich zu beschreiben. Dafür wurde bislang in Kauf genommen, dass diese Dichtefunktionale im asymptotischen Bereich, d.h. bei großen Abständen vom Mole-külen, nicht das korrekte Abfallverhalten zeigen. Sie fallen in der Regel viel zu schnell ab, was dazu führt, dass die damit erzielten Elektronendichten zu langsam abfallen. Während dies bei der Berechnung von Energetik und Struktur normal gebundener Moleküle leicht zu verschmer-zen ist, hat ein falsches asymptotisches Verhalten des Dichtefunktionals deutliche Auswirkun-gen auf molekulare elektrische Eigenschaften wie Multipolmomente und Polarisierbarkeiten. In jüngster Zeit interessiert man sich zunehmend dafür, asymptotisch korrekte Dichtefunktionale zu entwickeln. Dieses Interesse wird hauptsächlich (!) dadurch stimuliert, dass elektronische Anregungsenergien in der Dichtefunktionaltheorie auf dem Umwege über dynamische, also fre-quenzabhängige Polarisierbarkeiten berechnet werden können. Eine Untersuchung mit bereits existierenden asymptotisch korrekten Dichtefunktionalen zeigt allerdings, dass Verbesserungen in den dynamischen Polarisierbarkeiten nicht automatisch Verbesserungen bei den Multipol-momenten sowie den Elektronendichten bei großen Abständen zur Folge haben, selbst wenn hier ein prinzipieller Zusammenhang besteht. In der Praxis kommt es auf eine möglichst gute Balance zwischen Valenz- und asymptotischem Bereich an. Das Poster stellt ein neues, von uns entwickeltes asymptotisch korrektes Dichtefunktional vor, bei dessen Entwicklung das Augenmerk darauf gerichtet wurde, diese Balance zu wahren. Wir erläutern die Konstruktion des Funktionals, genannt PBE0-AC, und stellen die damit für eine Reihe von Testsystemen erzielten asymptotischen Elektronendichten, Multipolmomente und Polarisierbarkeiten vor. Die ermutigenden Resultate zeigen, dass mit dem neuen Funktional ein wichtiger Schritt zu einem praktisch anwendbaren und genügend genauen Dichtefunktional-Stö-rungstheorieverfahren für intermolekulare Wechselwirkungen gemacht wurde.

5.3.4 Heßelmann, Andreas; Jansen, Georg Calculating molecular electric properties and intermolecular interaction energies from density functional theory EMC2 Conference 'Exploring Modern Computational Chemistry', Nottingham/GB, 31.07-02.08. 2002

5.3.5 Heßelmann, Andreas; Jansen, Georg A Kohn-Sham-SAPT combination to describe intermolecular interactions 38.Symposium für Theoretische Chemie, Bremen, 25.8-29.8 2002

Abstract: This contribution presents a combination between Kohn-Sham Density Functional Theory and intermolecular Symmetry-Adapted Perturbation Theory (SAPT). While in the conventional SAPT the individual terms (namely the Coulomb, induction, dispersion and their corresponding exchange parts) are calculated via Many-Body Perturbation Theory our Kohn-Sham (KS) variant uses Kohn-Sham orbitals to describe the first-order intermolecular contributions and (time-dependent) coupled-perturbed KS theory to describe the static and frequency dependent response functions needed to calculate the induction and dispersion contributions. This scheme is potentially exact for the Coulomb, induction and dispersion contributions, i.e., these terms would be exact if the exact exchange-correlation potential and the exact response kernel of DFT were known. For the corresponding exchange kernel it is not potentially exact. Our results for a choice of small dimer systems suggest, however, that these terms are very well approximated. Thus, the KS-SAPT scheme seems to be a promising way to describe the intermolecular interactions between extended monomers. Moreover the dispersion energy problem in conventional supermolecular KS-schemes is naturally solved.

5.3.6 Marian, Christel M; Schneider, Frank; Kleinschmidt, Martin; Tatchen, Jörg Electronic Excitation and Singlet-Triplet-Coupling in Uracil, its Dimer, and Uracil-Water complexes

38th Symposium for Theoretical Chemistry, International University Bremen, 25.08-29.08.2002

Abstract: Electronic spectra of uracil in its diketo (lactam) form and the corresponding dimer have been investigated by means of the combined density functional and configuration interaction method (DFT-MRCI) of Grimme and Waletzke. In addition, model complexes of uracil and two, four, and six water molecules have been used to simulate spectral shifts due to hydrogen bonding in polar protic environments. Geometries of several low-lying excited electronic states have been optimized within the DFT/MRCI approach applying a numerical gradient procedure. Spin-orbit coupling has been determined for the correlated wave functions applying the one-center mean-field approximation. In accord with earlier findings, the first absorption Volume arises from the 1(π → π* ) S0 → S2 excitation. The excitation energy of this transition is considerably reduced by dimerization. On the other hand, complexation with water molecules causes a significant blue shift of n → π* excitations while leaving π → π* excitations nearly uninfluenced. The experimentally known weak phosphorescence originates from spin-orbit coupling of the π → π* excited T1 state and the n → π* excited S1 state possessing only a marginal oscillator strength.

5.3.7 Marian, Christel M.; Kleinschmidt, Martin; Tatchen, Jörg Electronic Excitation and Singlet-Triplet-Coupling in Organic Molecules: Application to Uracil. 12th European Seminar on Computational Methods in Quantum Chemistry Utrecht, The Netherlands, September 18-22, 2002

Abstract: With our new spin-orbit coupling kit (SPOCK), [1] spin-orbit matrix elements (SOMEs) for the combined density functional and configuration interaction method (DFT-MRCI) [2] or conventional Hartree-Fock (HF)/MRCI wavefunctions can be generated. Key features of the current version are I.) a fast determination of spin-coupling coefficients between configuration state functions (CSFs) for spin-dependent one-electron operators; II.) use of non-empirical atomic spin-orbit mean-field integrals [3]; III.) determination of spin-orbit related properties by quasi-degenerate perturbation theory. The application of these techniques and approximations in combination with the efficient determination of correlated wavefunctions and electronic energies in the DFT/MRCI approach paves the way for an economical but reliable description of spin-orbit effects in larger organic molecules. By means of the above mentioned methods, we have investigated the vertical and adiabatic electronic spectra of uracil as well as spin-orbit coupling between singlet and triplet states. In addition, the uracil dimer and uracil water complexes have been used to simulate spectral shifts caused by hydrogen bonding and solvent effects. In accord with earlier findings, the first absorption Volume arises from the 1(π → π*) S0 → S2 excitation. The excitation energy of this transition is considerably reduced by dimerization. On the other hand, complexation with water molecules causes a significant blue shift of n → π* excitations while leaving π → π* excitations nearly uninfluenced. The experimentally known weak phosphorescence originates from spin-orbit coupling of the π → π* excited T1 state and the n → π* excited S1 state possessing only a marginal oscillator strength.[4] [1] siehe [6.1.6] [2] S. Grimme, M. Waletzke: J. Chem. Phys. 111 (1999) 5645. [3] B. Schimmelpfennig, AMFI, Stockholm's University [4] siehe [6.1.7]

5.3.8 Raub, Stephan Quantum Chemical Studies on HEM-Imidazole 38th Symposium for Theoretical Chemistry, Bremen, 25.08.2002 – 29.08.2002

Abstract: We have performed DFT calculations on a Fe-porphyrine complex (FeP) coordinated with methylthiolat and imidazole. This was done to model the binding of 4-phenyl-imidazole to the active site of Cytochrom P450 14-alpha-sterol-demethylase (CYP51). Energy profiles have been recorded for the bond distance, torsion angle and tilt angle of the imidazole ligand with respect to the FeP. Optimized bond lengths and angles are in good agreement with the experiment. We find an eclipsed conformation of the imidazole-FeP complex with a low barrier to free rotation. Our calculations show that the tilting of the ligand (as observed in the enzyme) increases the preference of imidazole-based inhibitors for CYPs in fungi over human CYPs.

5.3.9 Reinhardt, Silke; Marian, Christel M. The reaction of TSDE with CH3NH2: solvent effects and entropic contributions 38th Symposium for Theoretical Chemistry, Bremen, 25.08.-29.08.2002

Abstract: (Trichlorosilyl)(dichloroboryl)ethane (TSDE) and methylamine are molecular precursors for high-demand ceramics. They forma polymer that is pyrolytically transformed to an amorphous solid of composition Si2B2N5C4.[1] Since translational symmetry is lacking in the Si-B-N-C network, structure determination requires the support of computer simulations. We present the results of a combined quantum chemical and first principles molecular dynamics study of a first aminolysis at the boron side of TSDE. To shine light on this reaction, we have firstly investi-gated the energetic course of the gas-phase reaction. Here, the aminolysis proceeds in three steps: a CH3NH2 molecule forms initially an adduct and a hydrogen chloride molecule is elimi-nated to form a hydrogen bonded complex. Dissociation of this complex leads to the products. The influence of an excess of methylamine on the mechanism and energetic course has been studied by means of Car-Parrinello molecular dynamics (CPMD). The free energies have been obtained by thermodynamic integration, using the Blue Moon ensemble. In an excess of CH3NH2 the reaction mechanism changes. Instead of a four-cyclic transition state a salt-like intermediate occurs. The free-energy gain of this aminolysis is much smaller than the internal-energy gain. This fact explains why this reaction is slower than a mechanistically and energeti-cally comparable process − the first amminolysis of BCl3.[2] [1] H. Jüngermann, M. Jansen, Mat. Res. Innovat. 2 (1999) 200. [2] S. Reinhardt, C.M. Marian, I. Frank, Angew. Chem. 113 (2001) 3795.

5.3.10 Torheyden, Martin; Jansen, Georg Ein einfaches Slaterfunktionsmodell des Wassermoleküls zur Berechnung elektrostatischer Wechselwirkungen Symposium "Molekulare Erkennung", Essen, 07.-09. 03. 2002

Abstract: Die Effekte der Überlappung molekularer Ladungsverteilungen sind von großer Bedeutung bei der Beschreibung von intermolekularen Wechselwirkungen im Bereich kurzer und mittlerer Abstände zwischen den Molekülen. Voraussetzung für eine korrekte Berechnung von Über-lappungseffekten auf beispielsweise elektrostatische Wechselwirkungen zwischen den Mono-meren ist natürlich eine gute Beschreibung der Einelektronendichte. In vielen gängigen inter-molekularen Kraftfeldern wird allerdings darauf verzichtet. Stattdessen nähert man die Ladungsverteilungen grob mit Hilfe von Punktladungen an und sucht die Überlappungseffekte mit anderen Termen, etwa Lennard-Jones oder ähnlichen Potentialen implizit mit zu erfassen. Klarerweise verwischt in solchen Kraftfeldmodellen dadurch die Bedeutung des Begriffs des elektrostatischen Wechselwirkungsbeitrags. Auf diesem Poster wird ein elektrostatisches Modell des Wassermoleküls vorgestellt welches diesen Nachteil vermeidet. Da aus der Atomphysik bekannt ist, dass Slaterfunktionen, also exponentiell abfallende Funktionen, ideal zur Beschreibung des radialen Abfallverhaltens der Elektronendichte von Atomen geeignet sind, wurde hierzu eine Linearkombination von Slater-funktionen verwendet. Die Slaterfunktionen wurden an den unterschiedlichen Atomen des

Moleküls zentriert und ihre Koeffizienten und Exponenten aus einem Fit an präzisen ab initio-berechneten Elektronendichten bestimmt. Das erhaltene Modell aus sieben Slaterfunktionen gibt nicht nur das elektrostatische Potential aus der ab initio-Rechnung sehr gut wieder, sondern auch die elektrostatischen Wechselwir-kungsenergien zwischen zwei Wassermolekülen, wie sie für mehr als 400 Konfigurationen des Dimers mit hochgenauer ab initio-intermolekularer Störungstheorie erhalten wurden - zu einem minimalen Bruchteil des Rechenaufwands.

5.3.11 Torheyden, Martin; Jansen, Georg A simple Slater function model of the water molecule for the calculation of electrostatic interactions 38. Symposium für theoretische Chemie, Bremen, 25.-29. 08. 2002

Abstract: As a part of ongoing work on an accurate intermolecular potential energy surface of the water dimer, in this poster we investigate a representation of the electronic density of the water monomer as a linear combination of Slater functions centered at the atomic sites of the mole-cule. With this model it is hoped to achieve an adequate description of the charge overlap con-tributions to the first-order intermolecular electrostatic interaction energy. The current model of seven Slater functions performs reasonably well over the wide range of a relative monomer orientations considered here, and even excellently well for those geometries of the water dimer where the hydrogen donor atom of one water monomer points towards a lone pair region of the other, like in the absolute minimum geometry.

6 Publications 6.1 Papers - appeared in 2002 [6.1.1] Atanasov, M., Schönherr, T.: The Unique Spectroscopic Behaviour of the Fe(II)-nitroprusside: a DFT Study of the Vibronic Coupling in the Ground and in the Lowest Excited States. Journal of Molecular Structure (THEOCHEM), Volume 592, 79-93, 2002 [6.1.2] Gastreich, M., Reinhardt, S., Doerr, M., Marian, C.M.: Modeling Si/B/N/(C) Ceramic Materials. NIC-Symposium 2001, NIC-Serie Bd. 9, Seiten 121-134, H. Rollnik (Hrsg.) und D. Wolf (Hrsg.), Forschungszentrum Jülich: 2002. [6.1.3] Heßelmann, A., Jansen, G.: First-order intermolecular interaction energies from Kohn-Sham orbitals. Chemical Physics Letters, Volume 357, 464-470, 2002. [6.1.4] Heßelmann, A., Jansen, G.: Intermolecular induction and exchange-induction energies from coupled-perturbed Kohn-Sham density functional theory.

Chemical Physics Letters, Volume 362, 319-325, 2002. [6.1.5] Kleinschmidt, M., Fleig, T., Marian, C.M.: Kramers-Type splittings in the X 2Π and a4Σ− states of CH and CD calculated in a Hund's case (a) basis Journal of Molecular Spectroscopy, Volume 211, 179-188, 2002. [6.1.6] Kleinschmidt, M., Tatchen, J., Marian, C.M.: Spin-orbit coupling of DFT/MRCI wavefunctions: Method, test calculations, and application to thiophene. Journal of Computational Chemistry, Volume 23, 824-833, 2002. [6.1.7] Marian, C.M., Schneider, F., Kleinschmidt, M., Tatchen, J.: Electronic excitation and singlet-triplet coupling in uracil tautomers and uracil water complexes – A quantum chemical investigation. European Physical Journal D, Volume 20, 357-367, 2002. [6.1.8] Reinhardt, S., Gastreich, M., Marian, C.M.: Quantum Chemical Investigation of Initial Reactions between the Molecular Precursor TADB and Ammonia. I: Gas-Phase reactions. Journal of Physical Chemistry A, Volume 106(16), 4205-4216, 2002. [6.1.9] Rosellen, U., Schmidtke, H.-H.: Optical Investigations of Alkali Metal Thiomanganates(II) Containing Isolated Complexes as well as Chain and Planar Structures. Inorganic Chemistry, Volume 41, 856-863, 2002. [6.1.10] Schlenz H., Kirfel, A., Schulmeister, K., Wartner, N., Mader, W., Raberg, W., Wandelt, K., Oligschleger, C., Bender, S., Franke, R., Hormes, J., Hoffbauer, W., Lansmann, V., Jansen, M., Zotov, N., Marian, C.M., Putz, H., Neuefeind, J.: Structure and analysis of Ba-silicate glasses: A collaborative study. Journal Non-Crystalline Solids, Volume 297, 37-54, 2002. [6.1.11] Schönherr, T., Linder, R., Rosellen, U., Schmid, V.: Spectroscopic and quantum chemical study on electronic and geometric properties of free and embedded dithizone molecules. International Journal of Quantum Chemistry, Volume 86, 90-99, 2002. 6.2 Papers - appeared in 2003 [6.2.1] Heßelmann, A., G. Jansen, G.:

Intermolecular dispersion energies from time-dependent density functional theory. Chem. Phys. Letters 367 (2003) 778 [6.2.2] Marian, C.M., Perić, M., Engels, B., Urban, W., Brown, J.M.: Spin-orbit and vibronic coupling effects in open-shell molecules: The link between theory and experiment. In: Interaction in Molecules - Electronic and Steric Effects, Hrg. S.D. Peyerimhoff, Wiley-VCH, Weinheim, 132-192, 2003. [6.2.3] Perić, M., Mladenović, M., Tomić, K., Marian, C.M.: Ab initio study of the vibronic and spin-orbit structure in the X 2Π electronic state of CCCH. Journal of Chemical Physics, Volume 118, 4444-4451, 2003. [6.2.4] Tatchen, J., Kleinschmidt, M., Marian, C.M., Parac, M., Grimme, S.: Quantum chemical investigation of spin-forbidden transitions in dithiosuccinimide. Zeitschrift für Physikalische Chemie, Volume 217, 205-230, 2003. 6.3 Papers - submitted in 2002 [6.3.1] Adamsky, H., Atanasov, M., Schönherr, T.: 1.98 The AOMX program. Comprehensive Coordination Chemistry - II; Section VII. Computational methods, Modeling and Simulation, Hrsg.: A.B.P. Lever, Elsevier, in press. [6.3.2] Gastreich, M., Gale, J.D., Marian, C.M.: Charged Particle Potential for Boron Nitrides, Silicon Nitrides, and Borosilazane Ceramics: Derivation of Parameters and Probing of Capabilities. Physical Reviews B, angenommen. [6.3.3] Kläui, W., Berghahn, M., Frank, W., Reiß, G.J., Schönherr, T., Rheinwald, G., Lang, H.: Tris(pyrazolyl)methanesulfonates: More Than Just Analogues of Tris(pyrazolyl)borate Ligands; N,N,N-, N,N,O-, and Other Coordination Modes. European Journal of Inorganic Chemistry, in press. [6.3.4] Schönherr, T., Atanasov, M., Adamsky, H.: 1.82 The Angular Overlap Model. Comprehensive Coordination Chemistry - II; Section VII. Computational methods, Modeling and Simulation, Ed.: A.B.P. Lever, Elsevier, in press. 6.4 Editorial work

[6.4.1] Schmidtke, Hans-Herbert Editorial Advisory Board, Spectrochimica Acta - Part A. [6.4.2] Schönherr, Thomas Guest editor for Structure and Bonding (Springer). 7 External collaborations

1. Dr. Joannis Apostolakis, LMU München

2. Prof. Michael Atanasov, Universität Marburg

3. Prof. Claude Daul, Universtät Fribourg, Schweiz

4. Dr. Julian Gale, Imperial College, London

5. Prof. Fritz Grein, Universität New Brunswick, Kanada

6. Prof. Stefan Grimme, Universität Münster

7. Prof. Dr. Georg Jansen, Essen

8. Prof. H.J. Aa Jensen, University of Southern Denmark, Odense, Dänemark

9. Dr. Andreas Kämper, MPI Informatik, Saarbrücken

10. Prof. V Kelloe, University of Bratislava, Slovakei

11. Prof. Miljenko Perić, Universität Belgrad, Jugoslawien

12. Prof. Pavel Peshev, Akademie der Wissenschaften, Sofia, Bulgarien

13. Prof. J. Olsen, University of Aarhus, Dänemark

14. Dr. Bernd Schimmelpfennig, Universität Stockholm, Schweden

15. Dr. L Visscher, Free University of Amsterdam, Niederlande

16. Prof. M Urban, University of Bratislava, Slovakei

17. Dr. Gerd Winter, FhG SCAI, St. Augustin