Psi-k - Ab initio (from electronic structure) calculation of complex processes in materials

The nature of hydrogen bonding and Density Functional Theory

From: joel ireta []
Sent: 14 August 2003 14:46
Subject: proposal for Ψk workshop for 2004

Dear Walter and Volker

Please find below our application for funding from the Ψk Network for holding a joint Ψk+Cecam workshop in 2004 in Lyon, France.

The aim of the proposed workshop is to discuss the nature of hydrogen bonding and the quality of exchange-correlation functionals to describe such interaction.

Hydrogen bonding (hb) is a prominent interaction in a wide range of materials (biomolecules, liquids, molecular crystals). It is well established that ab-initio techniques accounting for the electronic correlation are needed for an accurate description of hbs. Therefore, in recent years DFT has been more often used to study them. However controversy exists on the usefulness and reliability of DFT to accurately describe hb interactions. In this workshop we plan to concentrate on the accuracy of different exchange correlation functionals to describe hbs. We believe such kind of workshop is urgently needed. Thus participants will learn when and why DFT provides an accurate description of such important interaction.

Yours sincerely,

Joel Ireta
Martin Fuchs and
Matthias Scheffler

Title: Workshop on "The nature of hydrogen bonding and Density Functional Theory"

Location: "CECAM, Lyon, France"
Date: " June 2004 "

Organisers:
Joel Ireta (coordinator and corresponding organiser)
Email:
Phone: +49 (0) 30 8413-4818

Martin Fuchs
Email:

Matthias Scheffler
Email:

Fritz-Haber-Institut der Max-Planck-Gesellschaft
Faradayweg 4-6
14195 Berlin
Germany

1. Scientific Content
The hydrogen bond interaction (hb) is an attraction between an electronegative atom (donor), and an electron rich atom or region (acceptor) actuated by a hydrogen atom. Donor and acceptor atoms are commonly O, N, F, and C. According to its interaction energy the hbs can be classified as strong (> 0.6 eV), moderate (0.16 to 0.6), or weak (< 0.16 eV). The hbs are remarkably important for a broad range of materials; they are a key feature for understanding biomolecules, liquids, and molecular crystals. Consequently a huge number of experimental and theoretical works have been devoted to investigate this fundamental interaction. However controversies exist on its nature and strength. Measuring hbs is a challenge from the experimental and theoretical point of view. Regarding theoretical methods it is well established that ab-initio techniques accounting for the electronic correlation are needed for an accurate description of hbs, i. e. with an error bar of less than 0.04 eV (1 kcal/mol) for predicting the hb strength. Thus the observed underestimation of the hb strength by the Hartree-Fock (HF) calculations is overcome using correlated methods like second order Møller-Plesset perturbation theory (MP2), coupled cluster (CC), or configuration interaction (CI) methods. Also very large high quality basis sets to expand the wave function are needed to get reliable hb properties. This fact and the necessity of correlated methods to accurately describe hbs make such studies computationally too expensive and only applicable to molecular complexes of at most a few tens of atoms. Therefore strategies to study hbs with similar accuracy to MP2 or higher levels of theory but computationally less expensive are needed. In this vein, density functional theory (DFT) is a method that includes electronic correlation. Unfortunately the accuracy of DFT to describe the hb interaction relies on the functional used to approximate the electronic exchange and correlation (xc). However, some trends are known: i) the local-density approximation (LDA) overestimates the hb strength, ii) the generalized density gradient approximations (GGA) and hybrid functionals are more accurate to describe the hbs than LDA, iii) the most commonly used GGAs to treat hbs, the Becke exchange functional and the Lee-Yang-Parr functional for correlation (BLYP) and the Becke hybrid functional for exchange and the Lee-Yang-Parr functional for correlation (B3LYP) normally underestimate the hb strength. Other GGAs like Perdew-Wang (PW91), the Becke approximation for exchange and the Perdew expression for correlation (BP), or the Perdew, Burke and Ernzerhof (PBE) functional have been applied to study hbs but are not yet widely used. These functionals usually give similar or better results than BLYP and B3LYP. iv) The reliability of more recently developed xc-functionals that include the laplacian of the density and/or the kinetic energy density, is even less known. v) Standard xc approaches provide very low classical energy barriers for proton transfer between donor and aceptor atoms. Particularly GGA approximations grossly fail in systems where the transition state involve a rearrangement of sp2 bonds.

In the proposed workshop we are planning to bring together the key theoreticians who have analyzed the nature of hbs and the quality of xc-functionals for describing them. Also the participation of young scientists on the field will be encouraged. The duration of the workshop will be three days and will be restricted to a participation of 30 people. This workshop intends to be interdiciplinary due to the broad range of materials where hbs are important. Therefore we expect participation of scientists from biological, chemical, and physical fields.

The main objective of the proposed workshop is to discuss about the accuracy of different exchange correlation functionals to predict hb interactions; i.e. hb strength and geometry, hb cooperativity, phonon frequencies and proton transfer barriers in hydrogen bonding systems. We will encourage participation of theoreticians who have been applying DFT based methods to hydrogen-bonded systems or benchmarking xc functionals respect HF based methods (e.g.Quantum Monte Carlo, CI, CC or MP2).

2. Speakers

We plan to arrange lectures by invited experts, which will cover the more fundamental background on: exchange correlation functionals and its applicability to study long range interactions, old and new views on the nature of hbs, hb cooperativity, the role of the basis set for an accurate description of hbs with DFT and DTF description of dynamical properties of hydrogen bonding systems. The list of the speakers/participants may include:

3. Participants

The proposed workshop is mainly aimed at theoreticians who have been applying DFT based methods to hydrogen-bonded systems or benchmarking xc functionals respect HF based methods (e.g.Quantum Monte Carlo, CI, CC or MP2). Contributions from scientists working in biological, chemical and physical fields will be welcome. The total number of participants including speakers will be restricted to 30 people.

4. Budget.

The projected cost per participant, excluding expenses for traveling to Lyon, is about 250 Euro. This includes accommodation for three nights in Lyon, as well as meals. Our aim is to provide full support for invited speakers and participants. This gives a sum of 7,500 Euro. Also we plan to support traveling expenses for some speakers, which add 4,500 Euro to the budget. From the total, 12,000 Euro, we asked Ψk Network to contribute with 50%, i.e. 6,000 Euro.

5. Workshop location
We intend to hold the workshop in Lyon in 2004. This is easy to reach and the facilities provided by CECAM are ideally suited to holding such event.