The HECToR Service is now closed and has been superceded by ARCHER.

Microiterative QM/MM Optimisation for Materials Chemistry

The ChemShell computational chemistry environment may be used to combine an accurate (but expensive) quantum mechanical (QM) treatment of the active site with a more approximate, but cheaper, molecular mechanical (MM) treatment of the surrounding solid environment. ChemShell provides a means of integrating quantum mechanical (QM) and molecular mechanical (MM) software packages to perform combined QM/MM calculations. The external programs are used for energy and gradient calculations while ChemShell performs higher level tasks such as geometry optimisation.

In a standard geometry optimisation each step requires a QM (typically 100s of atoms) and MM (typically 1000s of atoms) evaluation at the new geometry. This means that the geometries of the QM and MM regions have to relax at the same rate. In a microiterative optimisation scheme, the system is divided into an inner region containing (at a minimum) the QM atoms, and an outer region containing the rest of the system. After each optimisation step of the inner region (the 'macroiterative' cycle), the outer region is fully optimised (the 'microiterative' cycle). By optimising in this way the number of QM evaluations is reduced significantly at the cost of increasing the number of MM evaluations of the outer region. As MM evaluations are usually much cheaper, this reduces the overall computational time which is usually dominated by the QM component.

The aim of this project is to improve the performance of ChemShell for large-scale geometry optimisation by implementing microiterative techniques in DL-FIND. This will be achieved by the following:

  • Implementation of the microiterative framework in DL-FIND by using similar techniques available in the existing HDLCOpt module in ChemShell. HDLCOpt was designed specifically for biological systems and is not suitable for materials chemistry. This work will bring microiterative functionality to DL-FIND, which has largely superseded HDLCOpt.
  • Extension of the microiterative framework in DL-FIND to support the partitioned rational function optimisation (P-RFO) and dimer transition state optimisation algorithms.
  • Development of a microiterative version of the nudged elastic band (NEB) method for reaction path optimisation in DL-FIND.

The overall outcome of this work may be summarised as follows:

  • Microiterative QM/MM optimisation methods were implemented for minimisation, transition state optimisation (P-RFO and dimer), and reaction path optimisation (NEB) in DL-FIND and ChemShell.
  • The algorithms have been tested on HECToR Phase 3 for the case of hydrogen dissociation over Li-doped MgO.
  • The number of optimisation steps required for convergence using microiterative (QM=macro cycles and MM=micro cycles) and standard minimisation algorithms was compared for a water sphere model with glycine solvated in water. A reduction of between 5 and 12 times the number of QM evaluations can now be achieved.
  • The microiterative optimisation was extended to two transition state optimisation algorithms: P-RFO and the dimer method. For the solvated glycine case there is an excellent agreement in optimised energies between the standard and microiterative runs.
  • For reaction path optimisation, the microiterative NEB algorithm was tested using solvated glycine.
  • The performance of the shell model microiterative optimisation algorithms was demonstrated on HECToR for correctness and performance, with a test system for the dissociation of hydrogen over Li-doped MgO. GAMESS-UK was used for the QM calculations and GULP for the MM calculations with a shell model polarised potential. The system consisted of a cluster of 6349 atoms plus 87 point charges around the edge of the cluster. The QM region consisted of 33 atoms in the centre of the cluster, including the hydrogen molecule, the lithium dopant, 5 oxygen atoms, and 25 magnesium atoms. There were 834 active atoms optimised in each run. For the microiterative calculations, the inner region was defined to be the same as the QM region (including boundary atoms). The results showed that all the microiterative algorithms are functioning correctly for shell model optimisation of materials systems, and that microiterative relaxation of the environment does reduce the number of macroiterations required.
  • In the particular case of the ionic embedding model, where the QM/MM boundary can be indistinct, the electrostatic potential (ESP) fitting procedure was found not to be robust. New research, beyond the scope of this project, is required to improve the ESP fit algorithm or else redefine the boundary so that there is a clear distinction between the QM and MM regions. For all other systems tested, the ESP fitting procedure works well and the microiterative algorithms are fully functional and ready for use.
  • These developments will be made available to users in ChemShell version 3.6, which is scheduled for release in summer 2013.

Please see PDF or HTML for a report which summarises this project.