ChemShell News and Highlights

Exascale materials modelling vacancy

Tuesday, 1 February, 2022

We are looking for a postdoc to join the Computational Chemistry Group at STFC Daresbury Laboratory to work on multiscale materials science workflows for future exascale computing platforms.

The 32-month position forms part of the PAX-HPC (Particles At eXascale on High Performance Computers) project funded under the UK’s ExCALIBUR initiative, which aims to research and develop software to meet the needs of the exascale era. The role will involve substantial research and development work in ChemShell, including implementation of advanced QM/MM approaches, underlying workflow improvements such as GPU-aware execution, and improved interoperability with leading QM codes for materials science, all working in close collaboration with our PAX-HPC partners. A second strand of work will target exascale I/O, to be showcased in several leading materials science packages.

For more information and to apply, please visit:

Closing date for applications: Monday 7 March 2022


Py-ChemShell 2021 released

Thursday, 23 December, 2021

We are delighted to announce the release of Py-ChemShell 2021 (v21.0), the third beta release of the Python-based version of ChemShell.

Py-ChemShell 2021 introduces biomolecular modelling functionality into Py-ChemShell, with a new guided workflow for protein setup, solvation and equilibration via an interface to NAMD, and automated import of CHARMM and AMBER forcefields for QM/MM calculations on biomolecular systems. Other new features include a periodic QM/MM embedding scheme for surface-adsorbate systems, and new interfaces to Molpro, Gaussian and MNDO.

As with the last release, Py-ChemShell 2021 is suitable for production calculations on materials systems and we would encourage materials modellers using the Tcl-based version of ChemShell to consider switching to Py-ChemShell for their production work if they haven't already done so. Tcl-ChemShell will continue to be maintained, but not actively developed in future.

We would also like to encourage experienced ChemShell users to try the new biomolecular modelling functionality and send us their feedback. For production calculations on biomolecular systems we continue to recommend Tcl-ChemShell for the time being, until the next beta release scheduled for summer 2022.

Py-ChemShell can be downloaded free of charge under the open source GNU LGPL v3 licence from this site. If you have any questions about the code, please get in touch via the user forums.


Geometry optimisation based on machine learning

Tuesday, 31 August, 2021

Johannes Kästner's group at the University of Stuttgart have extended DL-FIND to use Gaussian-process regression (GPR) to search for minima, transition states, and reaction paths.

Energies and gradients during the optimisation trajectory enter the GPR algorithms, which build up a surrogate surface that aids the optimisation. The resulting algorithm requires fewer steps than traditional approaches, especially for transition states and reaction paths. Moreover, Cartesian coordinates and, recently, also internal coordinates can be used.

The GPR-enabled version of DL-FIND is freely available.


A. Denzel and J. Kästner, "Gaussian process regression for geometry optimization" J. Chem. Phys., 2018, 148, 094114.

A. Denzel and J. Kästner, "Gaussian Process Regression for Transition State Search" J. Chem. Theory Comput., 2018, 14, 5777-5786.

D. Born and J. Kästner, "Geometry Optimization in Internal Coordinates Based on Gaussian Process Regression: Comparison of Two Approaches" J. Chem. Theory Comput., 2021. DOI: 10.1021/acs.jctc.1c00517



Py-ChemShell 2020 released

Friday, 3 July, 2020

We are delighted to announce the release of Py-ChemShell 2020 (v20.0), the second beta release of the Python-based version of ChemShell.

Py-ChemShell 2020 includes improved cluster cutting functionality, an interface to FHI-aims, improved wavefunction guesses for NWChem and a new set of tutorials covering first steps with the software up to full QM/MM calculations, as well as a huge number of improvements behind the scenes.

We now recommend that materials modellers still using the Tcl-based version of ChemShell should consider switching to Py-ChemShell for their production work. Tcl-ChemShell will continue to be maintained, but not actively developed in future. Biomolecular modellers should continue to use Tcl-ChemShell for the time being, but we expect to include biomolecular workflows in our next release due later this year.

Py-ChemShell can be downloaded free of charge under the open source GNU LGPL v3 licence from this site. If you have any questions about the code, please get in touch via the user forums.


QM/MM simulations help to resolve the sodium pumping mechanism of a light-driven ion transporter

Friday, 26 June, 2020

Krokinobacter eikastus rhodopsin 2 (KR2) is a light-driven sodium pump that actively transports small cations across cellular membranes. Such pumps are used by microbes to convert light into a membrane potential and have become useful optogenetic tools with applications in neuroscience.

An interdisciplinary consortium has applied time-resolved serial femtosecond crystallography to study the pumping process in KR2. The team of Dr. Jörg Standfuss (Paul Scherer Institute, Switzerland) has obtained high-resolution snapshots throughout the KR2 photocycle, which visualized sodium translocation over time. However, the sodium ion has the same number of electrons as a water molecule and is therefore difficult to differentiate in crystallographic data. Hence, Rajiv K. Kar and Igor Schapiro used the hybrid QM/MM method in ChemShell to assign the position to a sodium ion or a water molecule.

The QM/MM simulations both refined the structure and computed spectroscopic parameters. This allowed the precise assignment of either sodium or water to the experimental electron density. These calculations were instrumental to resolve transient binding sites of sodium and the involved key amino acids. These results provide a direct molecular insight into the dynamics of cation transport across biological membranes.


P. Skopintsev, D. Ehrenberg, T. Weinert, et al., "Femtosecond-to-millisecond structural changes in a light-driven sodium pump", Nature, 2020. DOI: 10.1038/s41586-020-2307-8


QM/MM studies of molybdenum and vanadium nitrogenases

Tuesday, 28 January, 2020

The iron-molybdenum (FeMoco) and iron-vanadium cofactors (FeVco) in the molybdenum/vanadium nitrogenase enzymes catalyze the reduction of dinitrogen at ambient temperature and pressure to ammonia according to the reaction equation: N2 +8H+ 8e- +16MgATP→ H2 + 2NH3+16MgADP. How this complex 8-electron process is carried out by these enzymes is still a mystery.

In a series of studies we have explored by QM/MM calculations, all performed with ChemShell, the resting state geometric and electronic structure of FeMoco in MoFe protein [1], the identify of the ligand in ligand-bound FeVco [2] and models for the N2-binding E4 state of FeMoco [3]. QM/MM methodology was shown to be critical for an accurate resting state FeMoco structure and to uncover the nature of ligand X in the structure of a recent ligand-bound iron-vanadium cofactor, FeVco in vanadium nitrogenase. Most recently, we have put forward a new model for the 4-electron reduced E4 state of FeMoco. Our model contains 2 hydrides that bridge Fe atoms no. 2 and 6 with a terminal sulfhydryl group and features favorable N2 binding and a subsequent reductive elimination step.


[2] B. Benediktsson, A. Th. Thorhallsson, and R. Bjornsson, "QM/MM calculations reveal a bridging hydroxo group in a vanadium nitrogenase crystal structure" Chem. Comm., 2018, 54, 7310-7313.

[3] A. Th. Thorhallsson, B. Benediktsson, R. Bjornsson, "A model for dinitrogen binding in the E4 state of nitrogenase" Chem. Sci., 2019, 10, 11110-11124.


Defects in the wide gap semiconductor GaN

Thursday, 7 November, 2019

GaN is a wide gap semiconductor that is a crucial component of blue light emitting diodes (LEDs), which are essential for solid state lighting. Both n- and p-type layers are required for LEDs, but achieving p-type GaN is very difficult; the only successful method is to very heavily dope GaN with Mg. Why no other dopants result in significant hole concentrations is not well understood. Moreover, there are many photoluminescence peaks observed from a broad range of differently prepared GaN samples whose origin is a major source of debate. Using the Chemshell code to perform hybrid QM/MM embedded cluster calculations, we have shown that Mg on a Ga site cannot act as a shallow acceptor in GaN and that holes are thermodynamically unstable, indicating that defect complexes and non-equilibrium processes dominate in p-type GaN. We have developed a new methodology to model shallow donors in GaN and have carried out a comprehensive analysis of native point defects in the material, indicating which defects could be the origin of the observed photoluminescence peaks. Our results have clarified the source of many experimental observations and have stimulated debate within the community on the defect properties of this important wide gap material.


Z. Xie et al., "Demonstration of the donor characteristics of Si and O defects in GaN using hybrid QM/MM", Phys. Status Solidi A, 2017, 214, 1600445.

Z. Xie et al., "Donor and acceptor characteristics of native point defects in GaN", J. Phys. D: Appl. Phys., 2019, 52, 335104.


In memory of Walter Thiel

Wednesday, 18 September, 2019

We were deeply shocked and saddened to hear that our colleague, friend and mentor Prof. Walter Thiel passed away last month. Walter Thiel made important contributions to many areas of computational chemistry over his long career, but ChemShell users will probably know him best from his crucial role in the development and application of combined quantum mechanical/molecular mechanical (QM/MM) methods to problems in biomolecular modelling. In addition to developing many of the underlying models which accurately describe such systems, he was a key developer of the ChemShell QM/MM package, beginning in 1998 with the EU-funded QUASI collaboration which introduced a range of new features into the program targeting enzymatic reactions. His semi-empirical electronic structure package MNDO was integrated with ChemShell and became a highly successful workhorse for enzyme active sites in the context of QM/MM. He subsequently pushed forward QM/MM methodology into areas such as molecular dynamics calculations, free-energy methods, microiterative optimisation techniques, solvent boundary potentials and methods for excited state optimisation and dynamics. His group applied these methods to a wide range of biomolecules, including of particular note a long series of investigations into the cytochrome P450 family of enzymes. Many other research groups followed his lead, resulting in hundreds of research publications which would not have existed without his pioneering work.

Walter Thiel continued to make major new contributions right up until his recent retirement from the MPI für Kohlenforschung in Mülheim, with recent method developments involving fully polarisable QM/MM models of enzymes, triple resolution QM/MM/coarse grained simulations, and QM/MM with periodic boundary conditions. His influence will continue to be felt for many years to come through the postgraduate students and postdoctoral researchers he mentored in Mülheim who have gone on to establish their own research programmes. We will also remember the special bond between the MPI and STFC Daresbury Laboratory, with several past and current Daresbury researchers having previously worked in his group. He was a wonderfully supportive, kind and inspiring supervisor and collaborator and will be greatly missed by us all.

Determining the energy band alignment between different TiO2 polymorphs

Tuesday, 25 June, 2019

TiO2 is a key material for photocatalytic water splitting, where it has been found that samples composed of mixtures of the anatase and rutile polymorphs outperform the pure phase samples. Several explanations for this observation were proposed with no consensus being reached until, using the ChemShell code to perform hybrid QM/MM embedded cluster calculations, we showed that it is a consequence of the fundamental band alignment between the two phases. Because of the relative positions of the valence and conduction band edges of both polymorphs, the effective band gap is reduced and charge separation occurs on photoexcitation that favours strongly the photocatalytic water splitting process. In a follow up study, we analysed the band alignment among the eight known polymorphs of TiO2, finding that the band edges can vary surprisingly widely, simply as a consequence of the difference in crystal structure among the phases. Our results helped explain many experimental observations that have consequences for energy storage, photocatalysis and optoelectronic applications.


D. O. Scanlon et al., "Band alignment of rutile and anatase TiO2", Nature Materials, 2013, 12, 798; J. Buckeridge et al., "Polymorph Engineering of TiO2: Demonstrating How Absolute Reference Potentials Are Determined by Local Coordination", Chem. Mater., 2015, 27, 3844.


Nitrite binding in three-domain haem-Cu nitrite reductase

Tuesday, 23 April, 2019

Copper-containing nitrite reductase enzymes (CuNiRs) play a key role in the global nitrogen cycle by reducing nitrite (NO2) to nitric oxide (NO). CuNiRs come in two-domain and three-domain forms, where the former have one cupredoxin domain and the latter have an additional cupredoxin or haem c domain. In two-domain CuNiRs, nitrite binding at the “T2Cu” active site is observed crystallographically, but in three-domain RpNiR it is experimentally elusive. It is hypothesized that a tyrosine residue from the linker part of the haem c domain blocks the substrate access channel and impedes nitrite binding.

With no experimental structures of substrate binding available we have used ChemShell QM/MM calculations to predict nitrite binding to native RpNiR. Our calculations suggest that NO2 binds to the T2Cu site in its oxidized Cu(II) state, but this occurs in an asymmetric L-shaped orientation via the nitrogen atom instead of the “top-hat” symmetric bidentate binding via the oxygen atoms observed in the corresponding two-domain CuNiRs. This observation is corroborated by a recent crystal structure of nitrite bound to mutant RpNiR.


K. Sen, M. A. Hough, R. W. Strange, C. W. Yong, and T. W. Keal, "A QM/MM Study of Nitrite Binding Modes in a Three-Domain Heme-Cu Nitrite Reductase", Molecules, 2018, 23, 2997.