Professor Catharine Esterhuysen
My current and on-going research focus is the analysis of weak non-covalent interactions between molecules and ions in the crystalline solid state, in order to understand their origin and to eventually harness them in the design of novel materials with interesting properties, especially porosity. In particular, I concentrate on unusual interactions, especially those that are primarily due to dispersion effects, such as aurophilic (Au···Au) interactions, as well as halogen bonds. In addition, the ubiquitous hydrogen bond cannot be ignored, although I generally focus on unusual aspects of hydrogen bonding, such as gold as a hydrogen-bond acceptor. For these studies I utilise a combination of experimental crystal structures, through "data-mining" of the Cambridge Structural Database, and computational methods, in particular Density Functional Theory (DFT). A limited selection of my current research in these directions is described below.
Iodide ions have a strong affinity to iodine molecules, forming polyiodide species such as triiodide (I3-), where the I–I2 interaction is extremely strong (~180 kJ mol-1). These polyiodide species have numerous applications, forming the blue colour of the starch-iodine complex that is used as a test for iodine and also playing a role in donor-acceptor materials exhibiting high electrical conductivity. The investigation of polyiodide containing conducting polymers, obtained by doping polymers with iodide, led to A. J. Heeger, A. G. McDiarmid and H. Shirakawa being awarded the Nobel prize in 2000. Although I3-···I3- interactions have been noted, calculations have focused on the intramolecular interactions within I3-. We are therefore the first to study these interactions and have shown that the chemical environment is vital for the correct modelling of the interactions. The important aspect is the treatment of dispersion vs. electrostatic interactions, where we have proven that very few computational methods are successful, making the choice of computational method for studying these types of interactions critical.
Metal iodide complexes show similar, albeit weaker, interactions with iodine. In particular, late transition metal iodides (which may or may not include stabilising ligands) form the strongest interactions with I2. We have shown that formation of metal iodides by oxidative addition of I2 to metal complexes can result in the formation of Pt(II)-I···I-Pt(II) chains. We are currently studying similar I-M-I···I-M-I (M=transition metal) interactions in order to gain a deeper understanding of the nature of these interactions relative to those between triiodide species, and determine the role played by the central metal atom with the aim of designing materials with tunable electronic properties analogous to the polyiodide materials currently used in electronics, solar cells and other electronic devices.
Hydrogen bonding is the interaction between a positively-charged hydrogen atom (due to its involvement in a polar bond to an electronegative donor atom) and an electronegative acceptor atom. The latter is typically O, N or F, but weaker interactions are observed with other elements, particularly C. Transition metals do not usually form hydrogen bonds as they are generally electropositive elements, but we have shown that gold, the most electronegative transition metal, can act as a hydrogen bond acceptor, thus proving that in some complexes it can be classified as a Lewis base, opposite to its more usual behaviour as a Lewis acid.
One of the problems with studying hydrogen bonding experimentally is that within crystal structures determined from X-ray diffraction data the hydrogen atom positions are poorly determined, with systematic shortening of the covalent bonding distance to H atoms. In a number of crystallographic software packages, for example the Cambridge Structural Database, this is currently treated by normalising the hydrogen coordinates utilising neutron data. However, it has recently been shown that neutron normalisation of O–H···O hydrogen bonds does not take polarisation into account, and therefore does not correctly adjust for the systematic error. An alternative statistical, polynomial method has been proposed that shows better correlation with experimental neutron data.An alternative method for determining hydrogen atom coordinates utilises quantum mechanical methods, particularly density functional theory (DFT), to model the correct hydrogen atom positions. This method may, however, be computationally expensive, and requires technical knowledge regarding the utilisation of such techniques. This means that the use of quantum mechanical methods is often beyond the usual toolkit of skills of a crystallographer, who nevertheless requires that the hydrogen atom positions are known with a high level of accuracy. We have compared the statistical method and DFT calculations for obtaining the hydrogen atom positions within O–H···O hydrogen bonds, and shown that, as might be expected, the accuracy of the H atom positions is sensitive to the DFT method used, often resulting in a poorer fit to experimental neutron diffraction derived O–H and H···O distances than the statistical method. The O–H···O angle on the other hand is usually more accurately described by DFT methods than by the statistical method.
A further focus area is the study of intermolecular interactions as a means of stabilising porous materials. This is work performed in collaboration with Prof LJ Barbour, where we utilise computational methods to understand the nature of various porous materials.
In particular, our interest is in explaining sorption phenomena, such as the difference in sorption isotherms obtained with increasing pressure of CO2 and HCCH for a macrocyclic Cd complex. It was shown that two gases have different electrostatic properties, hence they are absorbed into the pores within the Cd complex in different ways, as shown in Fig. 1
The electrostatic properties of CO2 have also been shown to be responsible for the differing sorption profiles observed for a series of metal-organic frameworks, Fig. 2. Despite compound 2 containing larger channels than compound 1, CO2 sorbed into the channels remains within the channels at lower pressures than for compound 1.
This is due to the complementary electrostatic profile of the framework in 2, which results in a stronger attraction for CO2 than in 1 or 3 (Fig. 3). This has important implications for the separation and sequestration of CO2, an important greenhouse gas.