44262-AC6
Theoretical Investigations of the Nature of Pi-Pi Interactions
State-of-the-art theoretical methods have been used to probe the fundamental nature of noncovalent interactions involving aromatic groups. Such interactions are central to protein folding, drug docking, crystal packing of organics, self-assembly, etc. We aim to determine the fundamental rules which govern these interactions and, ultimately, to learn how they can be precisely controlled.
1. What is the fundamental nature of pi-pi interactions? What are their physical origins (electrostatics, dispersion, etc), their geometrical preferences, their strength, etc? To address this question, we have embarked on a thorough, benchmark-quality theoretical study of the benzene dimer and its derivatives. A feature article in J. Phys. Chem. A (Sinnokrot and Sherrill, 2006) explores these and other issues, particularly what level of theory is required to achieve reliable results for these systems. This very successful article led to additional invitations for Feature Articles or book chapters on this subject. A chapter of Annual Reviews in Computational Chemistry is in press, and we are currently working on new Feature Articles for Physical Chemistry Chemical Physics (surveying the literature on pi-pi substituent effects) and for Journal of Physical Chemistry A (on approximate methods for noncovalent interactions). We have also addressed associated questions about the nature of CH/pi interactions (Ringer et al, 2006). We accepted the challenge offered by Schweizer and Dunitz, who argue that the lattice energy of crystalline benzene is the next logical target for high-accuracy computations of prototype pi-pi systems. We published (Chem Eur J, 2008) very high-quality values [estimated complete basis set CCSD(T)] for the lattice energy of crystalline benzene. Our error analysis and comparison to experiment indicates that we have achieved "chemical accuracy" in our computations, which to our knowledge is the first time this has been accomplished for the lattice energy of any organic crystal.
2. How do substituents tune pi-pi interactions? Here we wish to know the basic rules of the game, which could be used in rational design of drugs or supramolecular architectures, or in crystal engineering. Building upon previous work in our laboratory, student Steve Arnstein recently published potential energy curves of substituted benzene dimers in parallel-displaced (slipped parallel) configurations. This was the cover of the May 21, 2008 issue of PCCP. New student Ed Hohenstein has examined heteroatom effects by looking at pyridine dimer and the benzene-pyridine complex (submitted).
3. Potential energy surfaces for pi-pi interactions. We are concerned not only with the minimum-energy configurations of pi-pi interactions, but also the general features of their potential energy surfaces. This is important because in actual, complex systems, the fundamental noncovalent interactions may not be able to achieve their "optimal" geometric configurations that would be observed in smaller gas phase complexes due to steric constraints or secondary interactions. In our work on substituted and heteroatom derivatives, and on approximate methods below, we have examined not just minima, but entire potential curves. This is particularly critical in assessing approximate methods such as new density functional models meant to model dispersion, because some of them fail at large distances.
4. Additivity of pi-pi interactions. We are also interested in how pi-pi interactions may be adjusted or polarized based on other nearby chemical groups (such as another pi system). We have published a study on benzene trimers and tetramers, and we have carefully examined the effect of three-body effects in our computations of crystalline benzene discussed above.
5. Testing approximate models for pi-pi and other noncovalent interactions. Recognizing that large-basis CCSD(T) computations are very hard to extend to larger systems, we have also been searching for approximate methods which might be applicable to larger systems while still yielding results close to our benchmark results. In one paper by Tait Takatani (PCCP, 2007), we examine Grimme's spin-component-scaled MP2 approach and related methods. We also consider the effect of local correlation approximations, which we find work well in general, but have surprising difficulties for the aug-cc-pVDZ basis set. Ongoing work also considers force-field methods as even less computationally expensive alternatives. An assessment of density fitting / resolution of the identity approximations will go in a JPC A Feature Article in preparation.
6. Do idealized models of noncovalent interactions play any role in actual complex systems? We studied the geometrical preferences of sulfur-pi interactions as they exist in the H2S-benzene model system using quantum mechanics. Remarkably, the same geometrical preferences in H2S-benzene are observed in our statistical analysis of SH/pi interactions in the Protein Data Bank (PDB), as published in Protein Science. This finding appears to have profound implications for understanding biomolecular structure.
