Ionic liquids (ILs) are salts with uncommonly low melting points that are formed by a combination of specific cations and anions; they display distinctive properties and can be used in a variety of applications. The working temperature range of an ionic liquid is set by the melting point and the boiling or decomposition point. In particular, the melting point (Tm) varies substantially between different ILs for reasons presently not fully understood, but which we explore herein. We show that the melting points of imidazolium ionic liquids can be decreased by about 100 K if an extended ionic and hydrogen-bond network is disrupted by localized interactions, which can also be hydrogen bonds. Evidence for the presence of ion–ion interactions through hydrogen bonds was reported by Dymek et al., Avent et al., and Elaiwi et al. some time ago. It is reasonable to assume that the interesting features of the melting points must be related to the formation of structures in the solid and the liquid phases of the ILs. Extended hydrogen-bond networks in the liquid phase were reported with possible implications for both the structure and solvent properties of the ILs. Dupont et al. described pure imidazolium ILs as hydrogenbonded polymeric supramolecules. Antonietti et al. suggested that these supramolecular solvent structures represent an interesting molecular basis of molecular recognition and self-organization processes. However, in all of these examples it is suggested that hydrogen bonds strengthen the structure of ILs leading to properties similar to those of molecular liquids. This idea is also the basis of most of the structure– property relations discussed in the literature including quantitative structure–property relationships (QSPR) methods to correlate the melting points of ILs based on “molecular descriptors” derived from quantum chemical calculations. Such empirical correlations suffer from the fact that large experimental data sets are required and that the statistical methods used are rather complex. In addition, no interpretation of these fundamental physical properties at the molecular level is provided. Krossing et al. have developed a simple predictive framework to calculate the melting point of a given ionic liquid based on lattice and solvation free energies. They showed that ILs are liquid under standard ambient conditions because the liquid state is thermodynamically favorable, owing to the large size and conformational flexibility of the ions involved. This leads to small lattice enthalpies and large entropy changes that favor the liquid state. For such studies substituted imidazolium, pyrrolidinium, pyridinium, and ammonium cations have been used along with fluorometalate, triflate, and bis(trifluoromethylsulfonyl)imide anions. Unfortunately, Krossing s results do not correlate with experimentally obtained melting points for protic ionic liquids (PILs) reported byMarkusson et al. The reason for the large deviations of the predicted from the experimental melting points is probably related to the general trend of increasing Tm with the increasing size of the anions. We do not intend to present another framework for predicting ionic liquid properties here. Instead we want to demonstrate that in addition to the large size and conformational flexibility of the ions, local defects such as directional hydrogen bonds can significantly decrease the melting points of ionic liquids. For eight imidazolium-based ionic liquids we show that these defects can increase their working temperature range by up to 100 K and thus expand the spectrum of potential applications. This was suggested previously by Fumino et al. based on spectroscopic measurements and DFT calculations on IL aggregates. They assumed that local and directional types of interactions present defects in the Coulomb system which may lower the melting points, viscosities, and enthalpies of vaporization. In contrast, based on quantum chemical calculations, Hunt claimed that an increase in the melting points and viscosities upon methylation at C(2) stem from reduced entropy. Noack et al. showed very recently that neither the “defect hypothesis” of Fumino et al. nor the “entropy hypothesis” of Hunt alone can explain the changes in the physicochemical properties. However, in all these studies the data base was not sufficiently large and other effects such as volume changes could not be excluded for the ILs under investigation. [*] Dipl.-Chem. C. Roth, Dr. K. Fumino, Dr. D. Paschek, Prof. Dr. R. Ludwig Universit t Rostock, Institut f r Chemie Abteilung f r Physikalische Chemie Dr. Lorenz Weg 1, 18059 Rostock (Germany) Fax: (+49)381-498-6517 E-mail: ralf.ludwig@uni-rostock.de
Read full abstract