Electronic Structure and Computational Modelling

Winter Semester 2024/2025

Final project

Please read the following carefully!

For the final projects we give you a small list from which to choose. Please review it and tell us which project
you choose. Some of the projects are more practical while others require a better discussion of the theory at
hand. Projects will not be repeated, i.e. one group per project, unless different software is used (e.g. you could
study a molecule with octopus and someone else with psi4). Of course you can also suggest a project that you
came up with, but this must be discussed with us to evaluate whether it is feasible within the context of the
class.
The evaluation itself will consist of a presentation of about 20 minutes, which will showcase the work of the
project. This will be followed by some questions both regarding the presentation and some of the associated
theory background.
We are currently flexible with the dates, so as soon as you have your other exams scheduled let us know to
fix a date. Any deviation from this plan will be communicated in advanced.

During your work:

• Prepare a guideline of what you will do (i.e. how to tackle the problem at hand). We recommend making
use of eduroam to search the literature for additional background.

• The work is expected to be independent to a certain degree (showing the skills developed throughout the
practical classes). However, when in doubt ask (better to ask a question than to present a wrong result)

• Keep your workflow organized (separated folders, well named things). This is better for eventual debug-
ging.

• Remember the GIGO principle: garbage in, garbage out (more often than not, a meaningless input will
lead to a meaningless result rather than a crash. . . )

• Be careful when writing inputs, use visualization software for double checking structures and be critical
of the results you get (e.g. if are modelling an alloy and get a formation energy of -5 eV then something
is clearly wrong. . . )

Before the presentation

• You should save and send the input files you generated for you work, with a small summary explaining to
what each file corresponds. This is to allow easier debugging. Do not upload output files.

As a general guideline, your presentation should cover:

• Motivation: briefly explain the problem at hand and why you are interested in it.

• Method: briefly explain the methods you used to obtain your results (e.g. DFT, HF, CI, etc.). Mention
the most relevant approximations/limitations of the method for the task at hand.

• Convergence: present clear plots/tables that illustrate eventual convergence procedures (e.g. plane-wave
cutoffs, k-point grids (keep them homogeneous!), grid spacing, basis set). Explain clearly your choices for
the final results.

• Results: This will vary depending on the problem you choose, but as a rule of thumb some of the following
quantities discussed in the classes should be included:

– optimized geometry (including pictures of your system)

– band structure/density of states
– density maps
– energies (e.g. formation/cohesive/binding/surface/...)
– potential energy surfaces

You could of course consider another quantity of your choice, as long as it is reasonable within your time
constraints.
• Discussion: comment on the validity of your results within the given approximations. Compare to existing
values in the literature.
• Conclusions: A short summary of the presentation, answering the questions raised in the introduction.
Give an outlook on open questions that you would like to answer in the future.

Please be careful when building your presentation

• Choose readable font and font size, also for equations.

• All graphs must have readable labels. All axes must have both labels and units. Line styles and line
colors must be distinguishable (avoid light colors, e.g. yellow).
• All data/figures from the literature must be accompanied with the corresponding reference. The references
should appear on the same slide they are used.
• Avoid full phrases/large text and crowded slides.
• Generally images are better representations than tables
• Practice your presentation (preferably with an audience of your colleagues)!
 List of Problems
Often in the past the evaluations were done individually, although we have tested the possibility of allowing
group work. This year we will again allow the opportunity of doing the work in pairs, meaning that for any
given topic the workload will likely need to be adequately resized (e.g. adding more systems to the study).
Notice that because of this uncertainty in the number of people, the projects as written may not reflect
the actual expected workload, which may need to be adjusted accordingly (e.g. fewer or more systems to be
studied). As such, we recommend you take a look at the topic and decide from there.
As we said before, we incentivate you to come up with a proposal for your project topic. You can either
choose an interesting paper or something you have worked before with. If that is the case, let us known what
you have in mind so that we can make sure it is adequate for the workload of the project.

Problem 1 This project investigates the size-extensivity problem in computational chemistry through a systematic
study of multiple non-interacting helium atoms using various quantum chemistry methods. The calcula-
tions will be performed using Psi4. The study begins by performing a careful Hartree-Fock calculation
on a single helium atom. You will need to conduct convergence testing with respect to the basis set,
starting with minimal basis sets and progressing through the Dunning correlation-consistent series (cc-
pVDZ, cc-pVTZ, cc-pVQZ, etc.). This will establish your reference for comparing correlation energies.
The core of the project involves calculating energies for systems containing increasing numbers of helium
atoms (from n=1 to n=10) at fixed large separations (at least 5 Å between atoms). For each system size,
perform calculations using both restricted Hartree-Fock (RHF) and configuration interaction with single
and double excitations (CISD). The large separation ensures that the atoms can be considered effectively
non-interacting. For each calculation, analyze and record, the total RHF energy, The CISD correlation
energy, the computational time, the number of CISD amplitudes that needed to be optimized, the energy
per atom (for both total and correlation energy). Create plots showing how these quantities scale with the
number of atoms. Pay particular attention to the correlation energy per atom and analyse its evolution
with the number of atoms. Compare these results with MP2 calculations on the same systems. Analyze
how the computational cost scales with system size for each method. Create plots of computation time
versus number of atoms and determine the scaling exponents. Compare these with theoretical expectati-
ons for each method. Compare your findings with theoretical expectations from quantum chemistry texts
and discuss the implications for choosing appropriate methods for different system sizes.

Problem 2 This project investigates the accuracy of different computational methods in predicting molecular ro-
tational spectra by comparing calculated inertia moments and rotational constants with experimental
microwave spectroscopy data. The calculations will be performed using Psi4. The study begins with a
careful selection of small to medium-sized molecules for which high-quality experimental microwave spec-
troscopy data is available. A good test set should include molecules with different bonding types and
molecular masses (e.g. set of molecules includes water, ammonia, methyl fluoride, methyl chloride, and
formaldehyde). For each molecule, perform geometry optimizations using different levels of theory. Start
with standard DFT functionals like PBE and BLYP, then proceed to hybrid functionals such as B3LYP,
and finally include meta-GGA functionals like SCAN or the Minessota members. Perform a systematic
study of the series of basis sets, using the correlation-consistent series (cc-pVDZ, cc-pVTZ, cc-pVQZ) and
its augmented versions to assess basis set convergence. For each optimized structure, calculate the princi-
pal moments of inertia and rotational constants A, B, and C. The analysis should focus on how different
computational approaches affect the accuracy of structural parameters, moments of inertia and rotational
constants. Calculate statistical measures of accuracy such as Mean Absolute Error and Root Mean Square
Error for each method. Look for systematic trends in errors with respect to molecular size or type. Study
the convergence of properties with respect to basis set size and evaluate the computational cost versus
accuracy trade-off for different methods. Compare your findings with literature benchmarks and discuss
both the successes and limitations of current computational methods for molecular spectroscopy.

Problem 3 This project explores the challenges in describing the F3 – anion using different computational methods,
demonstrating how various levels of theory predict different results for this apparently simple system. The
calculations will be performed using Psi4. Compute the potential energy curves for this molecule using
Hartree-Fock, MP2 and CCSD (be sensible in your grid choice...). Use correlation-consistent basis sets of
increasing size (cc-pVDZ, cc-pVTZ, cc-pVQZ) to ensure proper description of the anionic system. Since
you’re dealing with an anion, consider adding diffuse functions (aug-cc-pVXZ series) and analyze their
impact on the predicted properties. In parallel, investigate the performance of different DFT functionals.
Include standard GGA functionals (e.g. PBE, BLYP), hybrid functionals (e.g. B3LYP, PBE0), and
modern meta-GGA functionals (e.g. SCAN). For methods that predict a stable symmetric structure,
calculate the dissociation energy for the reaction F3 – to F2 +F – . Compare the predicted values with
experimental data. Examine the basis set dependence of your results by performing calculations with basis
sets of increasing size. Create plots showing the convergence of key properties (bond lengths, dissociation
 energy) with basis set size. The final analysis should present a clear picture of which theoretical methods
are capable of correctly describing this system and why simpler methods fail. Discuss the computational
cost versus accuracy trade-off for the different methods.

Problem 4 This project aims to study the accuracy of different computational methods in describing weak interactions
using noble gas systems as test cases. The work will be carried out using the Psi4 quantum chemistry
package. The project consists of three main components. The first part involves a basic comparison
of computational methods. You will study the dimers Ne2 , Ar2 , and Kr2 using several levels of theory.
These should include Hartree-Fock calculations, DFT functionals, and MP2 as a reference method. For
each system and method, you will calculate binding curves and compare the predicted equilibrium distances
and binding energies with experimental values. If the project is done in group, then you should also extend
the investigation by studying noble gas trimers (Ne3 , Ar3 ). You should examine both linear and triangular
geometries and analyze how different methods handle dispersion effects. Throughout the project, careful
attention must be paid to convergence studies with respect to basis sets, which is particularly crucial
for weak interactions. You will need to generate potential energy curves for all systems and compare
equilibrium distances and binding energies. You should also discuss the balance between computational
cost and accuracy for the different approaches. The final presentation should include a critical discussion
of which methods are appropriate for different applications involving weak interactions, based on both
accuracy and computational cost considerations.

Problem 5 This project investigates the electronic and structural properties of van der Waals heterostructures com-
posed of graphene and hexagonal boron nitride (hBN) using density functional theory. The work will be
performed using the GPAW. The study begins by optimizing the individual monolayers of graphene and
hBN. You will determine their lattice parameters and electronic properties, paying particular attention
to convergence with respect to k-point sampling and energy cutoff and vacuum size. This initial step
serves both as a validation of your computational setup and provides reference data for the subsequent
heterostructure calculations. For the heterostructure, you will investigate different stacking configurati-
ons. The primary configurations to study are: boron-centered (where carbon atoms are positioned above
boron atoms), nitrogen-centered (carbon above nitrogen), and hollow-centered (carbon above the center
of the hBN hexagon). For each configuration, perform full structural relaxation using both standard DFT
and a van der Waals corrected functional. Calculate and compare the geometries, binding energies, band
structures and density of states for each stacking configuration. Pay particular attention to the behavior
near the Dirac point of graphene and any modifications induced by the interaction with hBN. The study
should include an analysis of the role of van der Waals corrections by comparing structural and electronic
properties obtained with and without these corrections. This comparison will demonstrate the importance
of properly treating long-range interactions in layered materials. The convergence studies should be tho-
roughly documented, including convergence with respect to vacuum spacing between periodic images, as
this is crucial for 2D materials. The final analysis should include a discussion of the relative stability of
different stacking configurations, the magnitude of the interaction between layers, and any modifications
to the electronic properties of graphene induced by the hBN substrate.

Problem 6 This project investigates the structural stability, phase transitions, and electronic properties of the archety-
pal perovskite SrTiO3 using density functional theory. The calculations should be performed using GPAW.
The study begins with the high-symmetry cubic phase of SrTiO3 (https://next-gen.materialsproject.
org/materials/mp-5229/). You will first perform careful convergence testing of the total energy with
respect to plane-wave cutoff energy and k-point sampling density. Using the converged parameters, op-
timize the lattice parameter of the cubic structure and calculate its electronic band structure, density of
states, and compare the predicted band gap with experimental values using different functionals (LDA and
PBE). The next phase involves studying the phase transition where TiO6 octahedra rotate, leading to a
tetragonal structure (https://next-gen.materialsproject.org/materials/mp-4651). You will need
to optimize its structure, including both lattice parameters and internal atomic positions. Calculate the
energy difference between cubic and tetragonal phases to determine the relative stability (LDA and PBE).
For both phases, perform a detailed electronic structure analysis (i.e. DOS and bandstructure). Gene-
rate clear visualizations of the structures using appropriate software like ASE. Next, compute the energy
versus volume curves for both phases. For each phase, compute the total energy at different volumes by
optimizing the cell at fixed volume. From these curves, extract the equilibrium volume, bulk modulus,
and the relative stability of the phases. The crossing point of the energy-volume curves will indicate the
theoretical transition point between the phases. You should also calculate the equations of state by fitting
the energy-volume data to an appropriate equation (such as Birch-Murnaghan). The study should include
an investigation of the role of the exchange-correlation functional by comparing results obtained with
different approximations (e.g. LDA vs PBE). The final analysis should provide a comprehensive picture
of the structural and electronic properties of SrTiO3 , including the energetics of its phase transition and
 the performance of different DFT functionals in describing these properties.

Project 7 This project investigates the formation and impact of point defects in monolayer MoS2 using density
functional theory calculations. The work will be performed using GPAW. The study begins by optimizing
pristine monolayer MoS2. You will need to determine appropriate convergence parameters, particularly
the vacuum spacing needed to avoid interaction between periodic images, plane-wave energy cutoff, and
k-point sampling. Calculate the structural parameters, band structure, and density of states of the pristine
system as reference data. The core of the project involves studying three types of point defects: a sulfur
vacancy (VS), a molybdenum vacancy (VMo), and a sulfur substitution with an oxygen atom (OS) and a
sulfur substitution with an selenium atom (OSe). For each defect, you will construct supercells of different
sizes (e.g. 3x3, 4x4, 5x5, 6x6 and so on) to assess the impact of defect-defect interactions and ensure
proper convergence of formation energies. For each defect type, calculate the formation energy using
appropriate chemical potential references (we will provide these values in the future). The calculations
should be performed using PBE. The electronic structure analysis should include the calculation of defect
states within the band gap (via band structure and DOS). Analyze the structural relaxation around each
defect type, including changes in bond lengths and angles. Create clear visualizations showing the local
distortions induced by each defect.