Summary
The group's main objective is to treat large complicated systems, and processes with theoretical methods. In order to model the microscopic behavior of chemical processes in the condensed phase, i.e. to simulate laboratory conditions and natural processes as closely as possible, it is necessary to deal with different traditional theoretical chemistry methods, ranging from molecular dynamics MD simulations based on empirical pair potentials to first-principles quantum chemical (QC) methods.
First-principles molecular dynamics FPMD simulations (like Car-Parrinello CPMD and Born-Oppenheimer BOMD simulations) and the quantum cluster equilibrium (QCE) method are important bridges between the traditional theoretical methods. These novel methods share the advantage of inherently treating cooperative effects. Therefore, they are appropriate for the investigation of highly associated liquids. Furthermore, the quantum cluster equilibrium method allows the treatment of systems with electronic structure methods of high accuracy. Within this model, post Hartree-Fock or correlated methods other than density functional theory can be applied, because less computational resources are needed than in FPMD simulations.
In the framework of traditional and FPMD simulations as well as within statistical mechanics, new ideas are being developed and exercised in methodological studies. Such methodological developments comprise, for instance, the combination of explicit relativistic descriptions with first-principles molecular dynamics simulations, and the derivation of thermodynamic quantities within the quantum cluster equilibrium theory. Other research topics are local properties, such as averaged local dipole moments in first-principles simulations, and the estimation of local hydrogen bond energies in an extended network, as well as the validation of methods.
Applications involve quantum chemical as well as molecular dynamics calculations of chemically interesting systems. We were able to predict the existence of the first C2v-symmetrical barbaralane molecule. Furthermore we study the activation of small molecules at metal centers. With regard to the research of complicated liquids, we were able to shed some light on unresolved phenomena like cooperativity of intermolecular interactions, hydrophobic hydration and solvent effects. We also investigate neoteric solvents like ionic liquids (IL) which play an important role in green chemistry. Recently, we discovered how water can act as glue to bind an ionic liquid ion pair (a so-called solvent-enhanced ion pair), by favourable arrangement in the solvent shell interfaces.
Projects |
Ionic liquids (IL) |
Intermolecular forces in ILs
The complex nature of ionic liquids is related with the sophistication of the intermolecular interactions present in the system. The potentiality of ionic liquids for a great number of applications relies on a better understanding on the balance of the specific interactions present in the ionic liquid media. Electrostatics plays an important role, but dispersion interactions and hydrogen bonding are present as well in ionic liquids, and they are mainly responsible for the different behavior of ILs and typical salts. The role of dispersion interactions is a major topic on the research activities of the group. With the aid of quantum chemistry methods, based on Density Functional Theory, we evaluate the relative stability of different conformations of the ion pairs, when the dispersion correction proposed by Grimme is considered or not. This study was extended to the bulk phase of ionic liquids using ab initio molecular dynamics (AIMD) simulations. The presence of hydrogen bond interactions in ionic liquids is also one of the subjects of interest. Hydrogen bonding interactions in imidazolium based ionic liquid are reflected in the charge distribution on the ion pairs, that depends on the conformational arrangement of the specific ion pairs considered. They are shown as well from an analysis of the frontier (HOMO and LUMO) orbitals. Energy surfaces, reflecting the change in the donor proton bond, indicate that regular hydrogen bonds are indeed present in ionic liquids. AIMD provide hints on the hydrogen bond network of ionic liquids in the bulk phase and how this network is modified by the presence of impurities as water.
References 64, 66, 67, 69, 79, 80, 90, 91, 92, 115 and work in progress.
ILs and reaction, impurities and mixtures
A mixture of liquids can have very different physical properties than the pure components caused. We try to rationalize how a binary mixture of ILs behaves, elucidating its possible non-ideality and the properties that are more affected for this non-ideal behavior. Using AIMD, we perform a comprehensive study on a one-to-one mixture of 1-ethyl-3-methyl-imidazolium thiocyanate [C2mim][SCN] and 1-ethyl-3-methyl-imidazolium chloride [C2mim][Cl] and we compare it to molecular dynamics simulations of the pure ionic liquids. There are some indications for weak non-ideal behavior. One is the partial suppression of coordination of the thiocyanate anion via the sulfur atom to the most acidic hydrogen atom of the imidazolium cation, which is visible in the neighbor distribution function (the integral of the radial distribution function). Similar effect can be observed for the coordination of the thiocyanate anion via the nitrogen atom to the rear hydrogen atoms of the imidazolium ring.
The role of the ionic liquid as solvent is not only dissolving the reactants but it plays a mayor role, interacting in a specific way with the different elements involved in the chemical reaction. Some reactions obtain better yields when performed in ionic liquids. The Diels-Alder reaction is an interesting example because it is a key step in many syntheses used to prepare cyclic compounds. ILs with strong hydrogen bonding interactions between the cation and the anion are poor solvents for Diels-Alder due to competition between the anion and the hydrogen bond acceptor dienophile to form hydrogen bonds with the cation. This is a clear example where the elucidation of the morphology of the hydrogen bond network is a key factor towards understanding the behavior of ILs in a given application. The use of ab initio molecular dynamics methods coupled with techniques as metadynamics, that allow the computation of the free energy based on biasing the dynamics with a history-dependent potential defined in a space of a set of collective variables, are suitable techniques to explore chemicals reactions under realistic conditions. It is possible then to sample in an efficient way the energy landscape of a specific chemical reaction and the effect of the ionic liquid to stabilize the reactants and transition states. We have considered in a first step the ionic liquid 1-ethyl-3-methylimidazolium chloride [C2mim][Cl], and we initiate a set of AIMD simulations, in order to understand the influence of the ionic liquid in the chemical reaction.
References 93, 97 103 and work in progress.
Theoretical investigations of the structure of ionic liquids
In order to investigate the structural and dynamical properties of ionic liquids, we used two different simulation techniques, namely classical molecular dynamics simulations and ab initio molecular dynamics simulations. For classical molecular dynamics (MD) simulations the determination of reasonable atom charges is very important. In order to evaluate several concepts, we carried out MD simulations with three different charge sets: the commonly used restraint electron potential fit (RESP), the natural bond orbitals analysis (NBO) and the shared electron number (SEN) analysis. The latter two are quantum chemistry based methodologies. We observed large differences between the structural and dynamical properties (see below) of the ionic liquid 1-n-butyl-3-methylimidazolium bromide calculated from the three charge sets. The conformation of the anions and cations determined from the RESP set was in good agreement with conformations from other calculations carried out in our group [80]. Whereas, the conformations from both the other charge sets showed small (NBO) and large (SEN) differences, respectively. A deeper analysis of the properties of the RESP system showed, that ionic liquids are build up of ion cages rather than of separated ion pairs. [95] To gain a deeper insight in the influence of different charges on the hydrogen bond ability of ionic liquids, we carried out additional simulations in the liquid phase. We altered the total charge as well as the atomic charge of the most acidic hydrogen atom and its neighboring carbon atom. We observed, that the alteration of the total charge of the ions leave the structure almost unaffected, while the structure was largely influenced by the hydrogen bond ability. [115]
The investigations of 1-ethyl-3-methylimidazolium aluminum chloride from AIMD showed, that the favored coordination area of the anions is around the most acidic hydrogen atom of the cation. On average, two chloride anions share this position, which means that the hydrogen bond is rather a traditional week than a strong directional one. The aluminum chloride anions build up structural units consisting of four monomers. The reason is a lack of electrons on the aluminum anion in the monomer unit. This could be quantified by the calculation of the electron localization function. [57]
Theoretical investigations of the dynamics of ionic liquids
As mentioned before, the choice of the charge set is important for the dynamics as well. To gain a deeper insight, we calculated correlation times for several time correlation functions. We could show that the correlations times calculated from the RESP and the SEN system are in the same order of magnitude, whereas the dynamics in the NBO system are much slower. Furthermore, we used two different definitions for the ion pair dynamics. By comparing these dynamics it becomes clear that ionic liquids are build up in ion cages. Besides, we found dynamics on four different times scales, which is an evidence for dynamical heterogeneity. [95] We carried out additional calculations in order to get a more detailed insight of the influence of charges on the hydrogen bond ability, see above. For the dynamics, we found that the total charges of the ions show a large influence (the higher the total charge, the slower the dynamics), whereas variations in the hydrogen bond ability do only manifest if drastic changes are induced. [115] Another study about the influence of the hydrogen bonds formed by the most acidic hydrogen atom of the cation, is the comparison of the dynamics of 1,3- dimethylimidazolium chloride and 1,2,3-trimethylimidazolium chloride. We carried out MD simulations as well as static quantum chemical calculations for both ionic liquids. The unexpected increase of the melting point and of the viscosity if the most acidic hydrogen atom is replaced by a methyl group can be explained by the constrained movement possibilities of the anion. [67] The above discussed results are examples of our work on neat ionic liquids. Additionally, we investigated: 1-ethyl-3-methylimidazolium thiocyanate, 1-ethyl-3-methylimidazolium aluminiumchloride [49, 57, 106), 1,3-dimethylimidazolium chloride [66] and 1,2,3-trimethylimidazolium chloride [66] from AIMD and further studies on 1-n-butyl-3-methylimidazolium bromide [115, 95] from MD.
Why does the hydrogen bond elimination lead to a reduction in viscosities?
In imidazolium-based ILs ([Cnmim]) substitution of the acidic hydrogen with a methyl group results in the reduction of one possible hydrogen bond donor. The elimination of this hydrogen bond interaction could be expected to lead to a reduction in melting point and a decrease in viscosity, because the ions should now interact more loosely; however the opposite is observed experimentally. We carried out static and dynamics calculations in order to study this controversial behaviour. A single energy path curve was calculated for the path from the chloride anion around the imidazolium ring via the acidic proton resp. the methyl group position. The energy barrier for the change from one side of the imidazolium ring to the other was found to be below 10 kJ/mol for the lower melting [C1mim][Cl], while it was increased above 40 kJ/mol for [C1C1mim][Cl]. Other pathways where the anion changes the sides of the imidazolium plane with an energy barrier below 10 kJ/mol were excluded for [C1C1mim][Cl]. MD simulations were carried out on the systems [C2mim][Cl] and [C2C1mim][Cl]. In agreement with the static QC calculations the anion of [C2mim][Cl] was found to be more delocalized than the one of [C2C1mim][Cl] at 15 times the corresponding average density. Refs.: 5
Solvent effects |
Mixtures of Ionic Liquids with Water
As ionic liquids are hygroscopic compounds, there is always a small amount of water present in them. It is thus very important to understand the influence of this water impurity on the properties and behavior of the IL. Therefore, we investigated different systems of water/IL mixtures with the aid of ab initio molecular dynamics. Apart from studies of one ion pair of the ionic liquids [C2mim][Cl], [C2mim][SCN], and [C2mim][DCA] in water, we recently performed a systematic study on [C2mim][OAc]/water mixtures in four different mixing ratios, including the pure ionic liquid and pure water. Considering the dipole moments, we found that the [C2mim] cation is generally depolarizing water in its close proximity, which leads to a reduced water dipole moment, whereas the [OAc] anion is polarizing neighboring water molecules, resulting in higher dipole moments. Imidazolium-based cations are weak hydrogen bond donors and therefore only barely interacting with water. However, [OAc] is a strong hydrogen bond acceptor. The hydrogen bond formed between [OAc] and the most acidic ring proton H2 in [C2mim] is very strong, resulting in configurations in which this proton is almost abstracted by the acetate. This leads to the formation of carbenes, which is in agreement with recent experimental findings. The proton abstraction effect gets lowered with increasing water content, suggesting that the carbene formation should no longer take place when water is added.
References: 66, 103 107, 109 and work in progress.
Solvated Biomolecules |
It is well-known that water plays an important role in biological systems. The chemical reactivity of amino acids can alter drastically upon salvation in water. One of the most obvious changes is that the former neutral species becomes zwitterionic. This is reflected in a large change of the dipole moment. The structure of the solvation shell of different amino acids was extensively investigated with several experimental methods. They show some contradictory results. The experimental uncertainties give rise to a wealth of computational studies of amino acids and their solvation shells. In our contribution, we investigated the hydration of alanine in water with different approaches, namely a continuum model and a cluster ansatz based on an ab initio molecular dynamics trajectory. The aim was the validation of the two methodologies. Furthermore, the structure of the solvation shell should be studied in detail. Our system contained one alanine molecule and several explicit water molecules in case of the cluster ansatz. The zwitterionic species of alanine contains three groups with different polarities: the basic, deprotonated carboxyl group, the acidic, protonated amino group and the aliphatic methyl group. The structure of the water molecules in the solvation shell depends strongly on the neighboring alanine group. While the charged groups form hydrogen bonds with the surrounding water molecules, this is not the case for the methyl group. Consequently, the solvation shell for the latter case is much broader. The dipole moments extracted form both methods deviate largely. The one determined from the continuum model is 3 Debye too small. This is an important result considering that many studies on electronic properties of hydrated molecules rely on such continuum models. We found large differences in the average dipole moments of the water molecules depending on their position relative to the functional groups. Based on these results, we can estimate, that one needs two solvation shells around charged groups and one around uncharged group for calculations with a reasonably converged average dipole moment of alanine.
References: 62.
We investigate the solvation effects of water and [C2mim][OAc] on the protected dipeptide Ac-Gly-Pro-NH2, to analyze the influence of the cis/trans conformation and the corresponding cis/trans ratio of the dipeptide in the different solvents. Simulations were started from the cis- and trans-conformations of the dipeptide. We observe similar structures of water for both conformations but the solvation of [C2mim][OAc] differ significant for the cis-trans conformations. Additional calculations and analysis are in progress in order to gain more insight into this ionic liquid effect.
Hydrogen Bond |
Hydrogen bonding plays an important role in chemistry and biology, e.g. in supramolecular or template chemistry. Although hydrogen bonding is an old and well known concept, there is still no condusive way to detect and define hydrogen bonds. The most recent short and wide definition of hydrogen bonds was provided by the IUPAC: “The hydrogen bond is an attractive interaction between a hydrogen atom from a molecule or a molecular fragment X–H in which X is more electronegative than H, and an atom or a group of atoms in the same or a different molecule, in which there is evidence of bond formation.” Our research activities are focused on detecting hydrogen bonds out of quantum chemical calculations. We were able to improve the idea of mapping the strength of the hydrogen bond onto an easily accessible quantity, namely the two-center shared-electron number, by extending the number of investigated dimers [40]. We classify the hydrogen-bonded complexes according to their acceptor atoms, and approach that provides more accurate results. We extended this work by further increasing the number of dimers and by introducing new approaches to detect hydrogen bonds [92]. We were able to correlate the frequency shifts and the acceptor-proton distances with the hydrogen bond strength. These correlations were applied to the hydrogen bonds present in polypeptides and other systems, showing the reliability of our correlations. An even more challenging topic is the detection of hydrogen bonds in the liquid phase, e.g. in ionic liquids. Therefore, we explore the intermediate bonds between different ion pairs via static quantum chemical calculations [90]. Per charge analysis we were able to reveal partial hydrogen-bonding character for ion pairs, where hydrogen atoms are involved. Analysing the frontier orbitals of these ion pairs, the possibility of hydrogen bonds was shown. We pointed out that the hydrogen bond between counterions can also have a disturbing rather than stabilizing role in their interaction.
References 40, 90, 92 and work in progress
Catalysis |
With the aid of a catalyst it is possible to accelerate a chemical reaction which goes along together with reducing the activation barrier. Sometimes, employing a catalyst is the only route to the synthesis of essential compounds for chemical industry. One well-known example is the synthesis of ammonia from the Haber-Bosch process. With the aid of theoretical methods parts of the energy surface of a catalytic cycle can be computed. The aim is to provide for each elementary reaction the activation barrier and to identify the rate-determining step. From this it might be possible to judge whether a catalyst is suitable for the desired reaction. To investigate the catalytic cycle we use AIMD simulations [93, 112] as well as static quantum chemical methods. Especially from the application of thermodynamic integration and metadynamics the energy surface may be reconstructed. As preferred electronic structure methods mainly density functional theory (DFT) is selected. Static quantum chemical calculations provide an overview over the energy landscapes of a multitude of different catalysts. The time consuming AIMD simulations are well suited to study the desired reaction under consideration of explicit solvents. In our group catalyst are investigated for different kind of reactions [93, 96, 97, 110]. In a former project, we gained experience with the problem of nitrogen fixation from iron and molybdenum complexes in collaboration with the group of Prof. Reiher at the ETH Zürich [71, 63, 38]. Another project deals with the decomposition of ruthenium oxoester [110, 87]. Therefore temperature-dependent AIMD simulations were carried out to investigate different pathways of the decompositions. The most recent project is the palladium catalyzed activation of methane and carbon dioxide in collaboration with the group of Muñiz at the University of Tarragona in Spain. For several different palladium catalysts the complete catalytic cycle was computed. The energy landscape for a selected catalyst of two elementary reactions was studied with the aid of AIMD and the free energy techniques mentioned above [97].
Quantum-Cluster-Equilibrium (QCE) |
Treating electronic structure in condensed phases has been feasible for some time by means of ab initio molecular dynamics. The Quantum-Cluster-Equilibrium model is, however, another constitutionally different approach for the investigation of liquids. Interestingly, the models which are adopted for the investigation of fluid phases had originally been designed for gases. The basic concept includes different-sized particles, that are - for the first – only weakly interacting so that each of them may be treated quantum chemically. The particles' geometries, energies and vibrational frequencies obtained by this may now be combined with simple models (particle in a box, rigid rotor, harmonic oscillator) to result in an approximation to the one-particle partition function. Meeting the equilibrium condition and assuming the total number of monomers to remain constant at each time, one may set up a polynomial whose roots determine the population of each particle at a certain phase point. Another polynomial fixes the temperature- and pressure-dependent phase volume. Due to the mutual dependency of volume and cluster population an iterative scheme is applied to gain self-consistent quantities.
In order to make the approach suitable for dense phases two additional terms need to be
introduced. Firstly, an attractive interaction between the particles is modelled by a
meanfield ansatz. This term is inverse proportional to the phase volume and is scaled by
an empirical parameter. Another parameter determines the clusters' volume which is
estimated by the atoms' van der Waals radii. This excluded volume has to be subtracted
from the free volume of translation when evaluating the translational partition function. The
two parameters are chosen to yield an isobar that agrees best with the experimental
behavior. The result of a QCE calculation is the system's partition function from which a
number of thermodynamic data may be evaluated.
The program Peacemaker has been developed in our group for carrying out QCE
calculations. It is freely available www.uni-leipzig.de/~peacemaker. A detailed
introduction to the methodology, the implementation in the Peacemaker code as well as its
application in a case study can be found in ref. 100.A first application of the method refers to the evaluation of thermodynamic quantities of the liquid state. In the past there have been several studies dealing both with values for the liquid phase and with phase transition quantities like vaporization entropies that could all be calculated in good agreement with experimental data. [65]
Less obvious but no less meaningful are the information that may be gained concerning the structure of the liquid under investigation. Thereby it has already been possible to contribute to the ongoing discussion about the coordination number in liquid bulk water. [74, 75]
A great advantage of the method lies in the fact that quantum chemical calculations are carried out only once for each particle and can then be used for several QCE calculations in contrast to AIMD simulation techniques. This offers the opportunity to apply explicitly correlated methods like the coupled-cluster approach at elevated basis sets or even to extrapolate to the complete basis set limit as has been done in the case of liquid hydrogen fluoride. In this study the structure in the liquid phase has been intensively studied, too. [98, 99, 111]
The QCE method can be extended to treat binary mixtures. [114] In the model it is assumed that there are two non-decomposable particles the so-called monomers in contrast to pure substances where there is only one monomer. All further particles are now built up of these two monomers. The methodology for binary system looks similar to the equations for the pure systems. However, there are multidimensional polynomials. The total mass of the system is introduced as a conserved quantity in order to obtain a nonlinear system of equations providing monomer populations and therby all the other particles' populations. The iterative scheme to determine the self-consistent quantities is the same for the binary mixture. Another difference concerns the description of the exclusion volume which depends on both monomer volumes, now. The scaling with one parameter analogously adopted. The extension of the method was tested with the binary system of water and dimethyl sulfoxide. Therby it has already been possible to contribute something about the coordination pattern of the water DMSO clusters. Additionally, excess Gibbs free energy could be reproduced in good agreement with experimental data.
However, there are some shortcomings that come along with the strengths of the approach. The isolated particle ansatz is not able to portray the real liquid phase. The concept to precalculate the particles allowing for methods that are computationally demanding forbids, on the other hand, any dynamics so that transport properties or diffusion processes are in principle out of the scope of this model. As a matter of fact, the quality of thermodynamic properties suffers from approximations that are made in the evaluation of the partition functions and consequently, quantities that contain second or higher derivatives of the partition function can no longer be obtained with high accuracy. Finally, the usage of empirical parameters is necessary in this approach but not desirable so that there is an effort to circumvent the application of one or better both parameters.
Validation… |
…of DFT
The performance of several general gradient approximation, meta general gradient approximation, and hybrid functionals was tested against MP2 for IL systems. Two dispersion corrected approaches (addition of van der Waals forces by a 1/r6 term (Grimme) and employing a dispersion-corrected atom-center dispersion pseudopotential (Roethlisberger)) were studied. For the [C4mim]+ cation neglecting dispersion, different trends for structural stabilities were observed. Investigating several [C4mim][DCA] ion pairs showed a mean absolute deviation from the MP2 interaction energy of 35.7 kJ/mol for HF and up to 33.2 kJ/mol for the DFT methods. The dispersion-corrected methods reduce the mean absolute deviation to less than 10 kJ/mol. Comparing adducts of the investigated ion pair with Diels-Alder educts (cyclopentadiene and methylacrylate) led to similar energetic differences as for the ion pairs. Furthermore, large deviations in geometries for the intermolecular distances were found for the HF approach (mean absolute deviation: 190 pm) and DFT (mean absolute deviation up to 178 pm), while for the dispersion-corrected methods the mean absolute deviation was less than 50 pm. For Hartree–Fock and all functionals it was found that the most stable [C4mim][DCA] structure is one where the anion lies in front of the C2-H2 position in plane with the imidazolium ring instead of the MP2 favored structure (by −10.7 kJ/mol) where the anion is located above the imidazolium ring. This structure is by 3.7 kJ/mol less stable for HF and by 4.3 kJ/mol less stable for B3LYP. Thus, depending on the methodology either the in-plane imidazolium ion pair or the above-plane imidazolium ion pair is favored. Contrary to this failure of DFT methods the dispersion-corrected DFT methods show the same trends as MP2. Refs.: 31
Thermodynamics |
In order to understand our complex systems from realistic quantities, we are also dealing with thermochemistry and thermodynamics. Partially, the thermodynamic properties are obtained with the aid of our Quantum Cluster Equilibrium (QCE) code PEACEMAKER [www.uni-leipzig.de/~peacemaker, 100], but also the calculation of thermodynamic state functions by means of static quantum chemical computations are carried out regularly [22, 33, 43, 60, 61, 63, 70, 71, 77, 78, 81, 87, 94, 113]. Additionally, thermodynamic quantities have been evaluated from (ab initio) molecular dynamics simulations [5, 7, 11, 97].
Most calculations are carried out in the frame of the so-called RRHO (rigid- rotor-harmonic-oscillator) model. The basic approximations of this model rely on the statistical thermodynamics of the non-interacting ideal gas, thus limiting its accurateness to low densities and high temperatures. The matter is even more urgent in the case of entropy computations, in which the largest contributions arise from low-lying vibrational degrees of freedom. These low- frequency vibrations are usually those which are most poorly described by a harmonic potential, i.e. RRHO-computed entropies are possibly no reliable estimate concerning the comparison to spectroscopically determined third-law entropies. Additional problems arise if RRHO entropy changes are computed for reactions involving a change in particle number during the reaction course. Such cases are of general importance e.g. in the field of homogeneous catalysis, where ligand association/dissociation reactions are accompanied by a change in particle number, but the phenomenon occurs as well in template- substrate associations and dissociations. During such reactions, a transformation of different degrees of freedom takes place, which in the frame of the RRHO model are treated by different degrees of approximation. Such computations often lead to quite large reaction entropies in the range of 40-50 J/(mol*K) at 300 K, showing little dependence on the actual system under investigation. Besides particle-number-conserving microsolvation approaches, the Quantum Cluster Equilibrium model provides an additional way to the thermodynamics of condensed phases in terms of ab initio calculations [36, 52, 114]. The approximative self-consistent QCE partition function allows a direct computation of thermodynamical quantities of pure and binary [114] phases over large temperature intervals such as densities and, as demonstrated in our group, entropies and heat capacities to a good precision. For liquid and gaseous water we showed several times the benefits of the QCE method in obtaining correct thermodynamic quantities [65, 74, 75, 100]. Furthermore, the evaluation of thermodynamic quantities with the QCE method can be applied to systems which are experimentally difficult to access, like HF [98, 99, 111] and AlCl3, and with lesser experimental data than in the case of water.
Supramolecular chemistry |
We are interested in the binding motifs of rotaxanes and pseudorotaxanes and their mimetics [47]. In general, rotaxanes consist of two mechanically linked subunits, a wheel and an axle, of which the latter penetrates the cavity of the wheel. To understand the binding motifs of these systems, we studied the binding energies and analysed the frequencies. We were able to identify a relation between the strength of an hydrogen bond and the charge of the acceptor oxygen atom and to explain substitution patterns. We investigated different minimums of the dethreading process of a pseudorotaxane [60]. Within this study, we tested different density functionals and we found large contributions (50 kJ/mol) to the binding energy. The thermochemical analysis revealed, that the two hydrogen bonds, found in the investigated pseudorotaxane, can be used as templates and behave as an entropy sink. The substitution pattterns on the binding between axle and wheel was also studied [61]. Substitutions with electron-withdrawing groups reduce the charge of the oxygen atoms and vice versa. The investigation of the effects of the solvent confirm the experimental assumption, that the systems are stable in CHCl3 and CH2Cl2, whereas in water the axle starts to dethread. Considering an exchange reaction, where the axle substitutes a solvent molecule inside of the cavity of the wheel, we could predict the free energy more accurate, than with a direct formation reaction or with an continuum solvation model. The frequency calculations revealed correlations of the C=O stretching modes and the N-H stretching modes to the substitutions and therefore to the individual hydrogen bond energies. A clear relation is observed between the NH stretch shifts and individual hydrogen bond energies [81]. In addition, we were able to relate the Hammatt parameter with these N-H stretch shifts. In future, we will study binding motifs of rotaxanes with asymmetrical wheels and longer axles and we will also take a look at the influence of neighbour axles.
References 47, 60, 61, 81
| ©2008 Kirchner Group Uni Leipzig |