Denna sida på svenska This page in English

Newest Doctors in Computational Chemistry

Linnea Lindh

Public Defence 6 October 2023

Title: Photophysics and Photochemistry of Iron Carbene Complexes


Nature captures sunlight via light-absorbing molecules.
Similarly, photosensitisers are used in applications of solar cells and artificial photosynthesis to absorb sunlight, and transfer the excited electron.
Successful photosensitisers have in the past been based on a Ru polipyridyl scaffold, despite Ru being one of the scarcest elements in Earth's crust.
This thesis work aims to replace Ru polipyridyl complexes by Fe carbene complexes, that by clever ligand design have approached suitable photosensitiser properties.

One crucial property that is not yet competitive for Fe carbene photosensitisers is how long they stay in the excited state, i.e. their lifetime.
This is controlled by the deactivation pathways of the molecule, dictated by the excited state landscape.
Several Fe carbene photosensitisers were in this thesis investigated by spectroscopic and computational methods, to understand their deactivation pathways.
For the Fe(II) carbene complexes investigated, small changes in the ligand structure influenced both what excited state (charge-transfer or metal-centred) that was mainly populated and the lifetime of the state.
For the Fe(III) carbene complexes investigated, there was instead one dominating charge-transfer excited state that was rather unaffected by changes to the ligand.
Furthermore, for the Fe(II) complexes metal-centred states played a large role in the deactivation pathway but for the Fe(III) complexes this was not the case.
Also, one Co(III) carbene complex was investigated which displays remarkable long lifetime and emission from a metal-centred state.

As a first step towards application, the electron-transfer properties of some of the photosensitisers were investigated.
Fe(II) complexes with a push-pull design were able to transfer electrons to TiO2 in a solar cell configuration.
The solar cell performance was however limited by an ultrafast recombination reaction, that brought a majority of the transferred electrons back to the photosensitiser.
The Fe(III) complexes investigated had long enough lifetime to participate in electron transfer with other molecules in solution, if the concentration was high.
Furthermore, at very high concentrations of the photosensitiser a light-induced charge-disproportionation reaction outcompeted all other deactivation pathways.
In a heterogeneous catalysis configuration, this reaction could generate long-lived Fe(IV) species with the correct additives.
The thesis work thus provide fundamental insights to the early implementations of Fe carbene photosensitisers in applications, by resolving key electron-transfer processes on the ultrafast timescale.

Link: 349661_1_G5_Linnea L.pdf (

Vilhelm Ekberg

Public Defence 26 September 2023

Title: Free-energy studies of ligand-binding affinities


In drug discovery, it is of utmost importance to accurately calculate the free energies of binding ligands to various protein targets, such as enzymes and receptors. We have assessed and used computational tools for this aim, most of them based on molecular dynamics (MD) simulations. We mostly used molecular mechanics (MM) in order to model the protein—ligand interactions, which is more approximate than quantum-mechanical (QM) methods, but necessary to reduce the computational cost when doing calculations on protein—ligand systems, which often contain tens of thousand of atoms.
In one study of a large set of protein—ligand complexes, we tried to improve the free energies of binding by using MD simulations with QM-derived charges, which sometimes led to improved results, but not always. We also ran QM/MM simulations on casein-kinase 2 (CK2), where the ligand and a few surrounding residues were treated at the QM level, and the rest of the system at the MM level. However, those results were unsatisfying. Furthermore, it is important and challenging to accurately model the large entropic contribution to ligand-binding free energies. This entropy largely stems from the fluctuation of the protein and ligand. We tried to estimate this entropy with methods based on fluctuations of interaction energies. We also saw how a combination of theoretical and experimental methods can shed light on phenomena like entropy—entropy compensation and halogen bonding. Additionally, we compared how MD and grand-canonical Monte Carlo (GCMC) can be used to assess dynamics and thermodynamics of protein—ligand binding for both buried and solvent-exposed binding sites.

Link: vilhelms_avhandling.pdf (

Magne Torbjörnsson

Public Defence 28 April 2023

Title: Computational Studies of Metalloenzymes


Enzymes are involved in most reactions in nature. They are important both for the understanding of biological life and for reactions of industrial interest, e.g. in the production of artificial fertilizers, the production of biomass or biofuels. Enzymes with one or several metal ions are called metalloenzymes. In this thesis we study three different metalloenzymes, nitrogenase, lytic polysaccharide monooxygenase (LPMO) and particulate methane monooxygenase (pMMO) with computer simulations. We use mainly combined quantum mechanical and molecular mechanical (QM/MM) calculations with the density functional theory (DFT) method.

Nitrogenase is the only enzyme that can break the triple bond in nitrogen molecules, making nitrogen available for plant metabolism. Previous studies have shown that different DFT methods give widely different results for the relative energies of different structures of putative intermediates in the reaction mechanism of nitrogenase. Therefore, we have tried to calibrate DFT calculations in two different ways. First, we use experimental data of structures and reactions related to nitrogenase to see what DFT functionals give the most accurate results.

Our results indicate that BLYP, B97D and MN15 give the best results. Second, we have developed a small and simple model system of the active site of nitrogenase, [Fe(SH)4H]−.

This model still shows a large variation in DFT estimate of the energy difference between structures protonated on Fe or on S, 25–163 kJ/mol. We then use a series of advanced and accurate QM methods, including coupled- cluster, selected configuration-interaction and multiconfigurational perturbation theory methods to calibrate the DFT methods. With this model, M06 and B3LYP give the most accurate results.

The second studied enzyme is LPMO, which is a copper dependent enzyme used for the degradation of polysaccharides, such as cellulose and chitin. The mechanism of this enzyme is quite well known but it has an unusual methyl modification of one of the ligands. Our calculations suggest that this group may protect the enzyme from self-oxidation.

The third studied enzyme is pMMO. This enzyme can hydroxylate methane. This enzyme is hard to study experimentally and the nature and location of the active site is still controversial. We have studied the reactivity of three putative active sites, involving mononuclear copper sites, and have shown that they can support similar reactions. The CuC site gives the most favourable energetics.

Link: e_spik_ex_Magne.pdf (

Joel Creutzberg

Public Defence 15 December 2022

Title: Extending relativistic linear response theory to address solvent effects

Abstract: The central aim of this thesis is to derive, implement and test new methods to calculate various types of spectroscopies of compounds containing heavy elements in an aqueous environment. Methods that can target such systems have to consider the following:
(i) It is crucial to take relativistic effects into account.
(ii) Modeling of larger systems is expensive in quantum chemistry. Thus, cheaper options need to be considered
for the water solvent.
(iii) Methods to calculate electronic spectra have to be able to model electronic excitations properly.
(i) The relativistic effects can be obtained by solving the Dirac equation. This yields a four-component wave function, but methods based on only two-components have been developed in this thesis. (ii) Larger systems can be tackled by dividing them into a region that is treated by methods from electronic structure theory, and a larger environment that is treated classically as a collection of localized static multipole moments (charges, dipole moments, etc.). In most such hybrid schemes (called QM/MM) we only take into account how the static multipole moments in the environment influence the wave function in the QM region. In this thesis, however, we allow mutual polarization of the regions through the polarizable embedding (PE) model. (iii) We calculate excited state properties through linear response theory. This has been developed to work with a variety of approximate state wave functions and has been extended to a relativistic framework. Moreover, it has been combined with PE. Yet, regular linear response theory suffers from problems in non-resonant regions of spectra. For this, we consider a variant of linear response theory, called the complex polarization propagator. Here, the life-times of the excited states are included in the response equations. This allows the calculation of spectra in regions that are problematic in regular response theory.
In this thesis, we have devised a method that combines relativistic CPP within a polarizable embedding framework. We employ the method on light-activated platinum complexes with application in chemotheraphy. Here, both relativistic and solvent effects are crucial to model the excitation processes. Moreover, we also consider the calculation of electronic circular dichroism for chiral organic molecules that contain heavy elements like iodine


Ernst Dennis Larsson

Public Defence 9 December 2022

Title: Towards a multiconfigurational description of the electronic structure in solids

Abstract: Materials of ionic crystals are ubiquitous in industrial chemistry.
For example, materials such as cerium dioixde (CeO2) are used in both self-cleaning ovens and to clean exhaust fumes from cars.
Other materials, such as titanium dioixde (TiO2) has been used in the solar-cell industry.
So-called garnets are used in several lasers.

In common for all of these areas of application, is that they are dependant on the motion of the electrons in these materials.
In order to understand how electrons behave and interact, quantum mechanics is required.
A major problem that immediately arises when applying quantum mechanics to crystalline materials, is that crystals are, from a quantum mechanical perspective, enormous.
One single crystal can contain as many as Avogradro's number of atoms.
Quantum mechanical calculation are very demanding, with even the most approximate methods available today being limited to around 10 000 atoms.
The type of methods used in this thesis, generally known as wavefunction theory, are roughly limited to around 100 atoms, depending a bit on what part of the periodic table that is explored and what type of property that is studied.

Methods that fall within wavefunction theory have the advantage against more approximate methods that they follow a fairly strict ladder of increasing accuracy.
In other words, the predicted results can, in principle, be improved by choosing methods from higher up on the ladder.
Of course, the higher up on the ladder a method is, the more computationally expensive it is.
It is therefore not necessarily affordable to move enough steps on the ladder, such that the desired accuracy can be reached.
For that reason, there needs to be some form om compromise when modelling crystals -- in order to improve the description of the electronic structure, the atomic structure has to become more approximate.
Models of that kind are usually referred to as embedding methods.

The purpose of this thesis has been to develop an embedding method for crystalline ionic materials.
This was achieved by developing a computer code called SCEPIC, that generates so-called ab-inito model potentials.
As a part of this thesis work, this method was evaluated in order to provide guidance to other researchers on how to best apply this method.


Eric Fagerberg

Public Defence 7 October 2022

Title: Assessing the structural and dynamical properties of concentrated solutions of the disordered proteins Histatin 5 and its tandem repeat

Abstract: Intrinsically disordered proteins are distinguished by a lack of distinct three-dimensional structure, existing instead as an ensemble of heterogenous structures. In this research, the effect of crowding on these proteins is investigated using a combined approach of experiment and computer simulation, mainly using coarse-grained simulation models to make simulation computationally feasible at the high concentration conditions crowding is displayed.
Firstly, the saliva protein Histatin 5 (Hst5) is studied with SAXS, where a selection of coarse-grained models were evaluated using the SAXS data. It was determined that no model could provide adequate simulation-experiment agreement, but a best-performing model could be established. This model predicted moderate change in structure with crowding in the case of Histatin 5.
It was postulated the moderate effect of crowding on Histatin 5 was due to its short sequence-length. Thus, the dimer of Hst5 was formed and subjected to investigation by SAXS and computer simulation for crowding effects. The dimer was more challenging to model with a coarse-grained model, and circular dichroism data suggested secondary structures to be present, which a coarse-grained model cannot capture. Atomistic modelling followed, which however did not perform better than the coarse-grained models, showing the importance of further developing these models to represent intrinsically disordered proteins.
Atomistic modelling was also performed at high concentrations of Hst5 5, combined with quasi-elastic neutron spectroscopy to elucidate diffusion behaviour at crowded conditions. Diffusion decreased with increasing protein concentration, with temperature effects following Stokes-Einstein beha- viour and increses in salt content to decrease diffusion. Depending on assumptions on the relation between effective- and translational-diffusion, the atomistic model displayed semi-quantitative agreement with experiment.
Using neutral polymeric crowders rather than self-crowding showed no impact on structure, as investigated by SAXS. Using DLS did as well not reveal any crowding impact, with the exception of Ficoll®, where Hst5 seemed to modulate Ficoll® self-crowding behaviour in terms of diffusion, decreasing the self-crowding effect. Several coarse-grained models showed similar non-existant effects on structure by crowding, with small deviations from experiment.
Benchmarking three coarse-grained models indicate higher degree of finegraining and additional parameters does necessarily follow the intuitive notion of increasing performance, with the most advanced not having as good performance as the two simpler models in terms of predicting radius of gyration.


Sara Haddadi

Public Defence 10 June 2022

Title: Polymer-Mediated Interactions and Phase Behaviour of Polymer-Particle Dispersions

Abstract: Interactions between colloidal particles can be modelled by particles grafted with polymers. In this work, structural and physical properties of colloids are investigated under variation of parameters such as pH, ionic strength, and temperature, where aggregation and cluster formation can be monitored in aqueous solution. Being the subject of our work, in particular, we show that linear or polymer-like clusters can be formed if long-ranged repulsive barriers are combined with very short-ranged attractive minimums stimulating particles to form highly anisotropic structures. This is adjusted by changing the properties of particles and the dispersing medium. Besides, we utilize Metropolis Monte Carlo (MC) simulation to investigate the behavioural change of these particles with a focus on the types of clusters formed. A simplistic potential of mean force is adopted for the simulations, but we also invoke a more elaborate model, to demonstrate that similar interactions can be obtained when the grafted chains are treated explicitly. An important criterion in these studies is that the particle size is large enough to allow structural analyses via microscopy. The range of electrostatic interactions is adjusted by the ionic strength, and the strength of the short-ranged attraction is changed via hydrophobicity regulation of the grafted layer through temperature variation. The results revealed that highly anisotropic structures which resemble linear or branched polymers were the clusters at equilibrium. We could also investigate the effect of polymer addition to the particle dispersions. We could detect a non-monotonic temperature dependent aggregation of particles from attraction to repulsion to attraction, where the polymer-mediated interactions were repulsive. The results were validated against experiments.
The next phase of this work is devoted to the study on capillary induce phase transitions with an experimental focus on polymer solutions containing PNIPAM at the presence of hydrophobic surfaces (mesoporous silica) as a function of pH, temperature and chain length. The capillaries/confined geometries are known to influence the phase diagram of polymer solutions where condensation of bulk solutions may occur close to the surfaces. This work is performed using a combination of experiments and theories where a shift to the LCST (lower critical separation temperature) is presumed to occur, resulting in a capillary-induced decrease in the LCST. 


Justin Bergmann

Public Defence 3 June 2022

Titel: Computational protein crystallography : How to get the most out of your data.

Abstract: It is important to obtain accurate three dimensional structures of molecules and proteins to understand and predict their function and behaviour. X-ray crystallography is the major technique to determine three dimensional structures of proteins. Although there have been major improvements on the experimental side in determining crystallographic data, only small progress has been made on the computational side to get a correct model and
interpretation of this data.
In small-molecule crystallography, some of the shortcomings in the model have already been overcome, but in protein crystallography they still remain. Therefore, we have adapted the Hirshfeld atom refinement from small-molecule crystallography to make it available also to protein crystallography. This enables improved modelling of high-resolution protein data. To achieve this goal, we combined the molecular fractionation with conjugate caps approach with the Hirshfeld atom refinement. We call this combined method fragHAR. With fragHAR, we could perform the first Hirshfeld atom refinement of a metalloprotein.
Furthermore, we improved and applied the quantum refinement method, which employs quantum mechanics calculations to obtain a chemically and physically correct model for all parts of the protein, especially the active site. With quantum refinement, it is possible to distinguish between different interpretations of the structure, e.g. the elemental composition or the protonation state, even from medium-resolution crystallographic data. In this thesis, quantum refinement was improved for highly-charged systems by applying a continuum-solvent description of the surroundings in the quantum mechanics calculation. Furthermore, quantum refinement was applied to settle the nature of the unusual bidentate ligand in V-nitrogenase and the protonation state of the MoFe cluster in Mo-nitrogenase when inhibited by CO. For a recent structure of Mo-nitrogenase, we showed that there is no experimental support for the suggestion that N 2 is bound to the MoFe-cluster and presented a more likely model. We have also identified the most probable protonation states of the active site in acetylcholinesterase before and after inhibition by nerve agents. Finally, for triosephosphate isomerase we used a joint X-ray and neutron quantum refinement to investigate the hydrogen bond between an inhibitor and Lys-13.


Samuel Stenberg

Public Defence 25 May 2022

As a PhD student in theoretical chemistry, i developed theoretical models that describe charged particles close to conducting surfaces. Such systems are common in both biological and non-biological systems. Because of my interest in programming i got the opportunity to develop my own simulation software, which i used to study these systems. One particular system of interest was electrical double layer capacitors (EDLCs). In EDLCs an electrolyte is confined between conducting electrodes. Using the methods we developed we were able to simulate these systems, and study the behaviour of the electrolyte as a constant potential was applied between the electrodes.


The thing that i will remember most of my time here, is not the research. The thing that makes theoretical chemistry great is the people, and in this regard i think it is a unique place. No matter who you are, you will always find a friend at theoretical chemistry.

Vidar Aspelin

Public Defence 8 April 2022

The thesis, Modeling of inorganic ions in aqueous solution, focuses on the design and application of atomistic simulation models, or force fields, for inorganic ions.


The lack of a thermodynamically robust force field for the thiocyanate ion motivated the development of a new force field that could reproduce both bulk and interfacial properties for a wide range of salt concentrations. The new, carefully validated force field provides novel insight at the molecular scale into the experimentally observed differences between two thiocyanate salts, NaSCN and KSCN. The force field was further applied in a study of the consecutive binding of potassium ions to a ditopic receptor: a bis-crown ether analogue of Tröger’s base.  While a counterintuitive enthalpic stabilization of the binding of the second potassium was observed experimentally for all studied salts, KCl, KI, KSCN, and K2SO4, the enhanced stabilization observed for KSCN was studied in light of the weaker hydration of the thiocyanate ion compared to the other studied anions, resulting in an enrichment of thiocyanate ions close to the receptor.


A large part of the research was devoted to the development and implementation of methods in the simulation software Faunus, such as the q-potential for handling long-ranged electrostatic interactions, the preferential sampling method for improving sampling, and the charge move method to enable the direct displacement of charges. Combining these with the Wang-Landau algorithm resulted in a platform of methods for efficient calculation of solvation free energies of ions.