Library · Simulation & modelling
What Density Functional Theory & Quantum Chem got wrong, from 31 theses
Summary written by a model from the records below: Computational records demonstrate that density functional theory and quantum chemical methods frequently encounter convergence failures, excessive scaling costs, and inaccuracies across correlated materials and non-covalent complexes. Researchers often reject standard functionals, semi-empirical approximations, and uncorrected treatments when modeling reaction barriers, dispersion forces, and electronic excited states. These records come from graduate theses, 2021 to 2026, each labeled with its degree where the repository states it, and each links to its thesis. They were extracted by language models reading the full text, so treat each as a lead to read, not a verdict.
Plus 9 problems the authors left open, listed at the end.
Failure modes, by theses reporting them
Select a mode to jump to its records.
Density functional approximations fail to accurately capture band gaps and electronic states in correlated systems
Semi-local functionals systematically underestimate band gaps in insulating polymers and incorrectly predict metallic states for titania nanoparticles and correlated superlattices. Mean-field approaches, range-separated hybrids, and standard functionals also mischaracterize spin states or overbind without explicit correlation treatments like Hubbard U corrections.
Tried and failed
density functional theory with generalized gradient approximations applied to wide bandgap polymer electronic structures. Reason: Standard semi-local and van der Waals functionals systematically underestimate fundamental band gaps in insulating polymers.
First-principles simulation of charge injection at metal/polymer interfaces · Imperial · PhD
Tried and failed
density functional theory single particle calculations applied to correlated superlattice electronic ground state. Reason: Predicted metallic state with high density of states instead of the experimentally observed insulating gap
Periodically Modulated Electronic States in Natural Superlattices · MIT · PhD
Tried and failed
dielectric-dependent range-separated hybrid density functional theory applied to liquid water and hexagonal ice. Outcome: worse than baseline. Reason: Significant dielectric screening underestimation leads to severe electronic band gap overestimation.
Nonempirical hybrid functionals for advanced electronic-structure calculations · EPFL · PhD
Tried and failed
generalized gradient approximation density functional theory applied to rutile titania nanoparticles. Reason: incorrectly predicted metallic behavior instead of the true semiconducting electronic structure
Analyzing and improving electronic structure calculations of catalytic interfaces using density functional theory and machine learning · Georgia Tech · PhD
Tried and failed
Hartree-Fock mean-field theory applied to correlated electron metal-insulator transitions. Reason: neglect of quantum fluctuations artificially stabilizes the symmetry-broken insulating state over metallic state
Quantum phases of semiconductor moiré materials · UT Austin · PhD
Tried and failed
Standard GGA density functional theory calculations applied to correlated transition metal complex electronic structure. Reason: Failed to capture strong electron correlation in localized d-orbitals without Hubbard U correction
Study of high quality factor resonators · Cornell · PhD
Tried and failed
hybrid functional density functional theory calculations applied to semiconductor quantum dot ground-state spin. Outcome: did not generalise. Reason: calculations predicted singlet ground states, failing to reproduce literature-reported non-singlet states
Colloidal Semiconductor Nanocrystals: Tools For and Insights From First Principles Investigations · MIT · PhD
Density functional methods produce severe errors when predicting reaction barriers, redox potentials, and charge transfer energetics
Calculations for interfacial electron transfer, injection barriers, and redox reactions often miscalculate onset potentials or fail to represent intermolecular charge transfer distributions. Continuum solvation models and pair-density functional approximations also significantly overestimate reaction and torsional energy barriers.
Tried and failed
Time-dependent density functional theory applied to intermolecular charge transfer energetics. Reason: Failed to reliably capture configuration-dependent charge transfer distributions between donor and acceptor molecules
Computational Studies on Organic Framework Materials as Absorbants and Sensors · Cornell · PhD
Tried and failed
density functional theory on single chain models applied to predicting electron injection barriers. Reason: systematic numerical discrepancies up to 0.7 eV compared to experimental measurements
DESIGNING POLYMERS RESISTANT TO ELECTRIC FIELD EXTREMES WITH MATERIALS MODELING AND MACHINE LEARNING · Georgia Tech · PhD
Tried and failed
implicit continuum solvated density functional theory modeling applied to spinel oxide electrocatalytic reaction pathways. Reason: assumed reaction pathway produced unphysically high energy barriers and severely overestimated onset potential
The Oxygen Reduction Reaction on Spinel Oxides in Alkaline Media · Cornell · PhD
Tried and failed
Density functional theory redox potential calculation applied to transition metal oxide redox transitions. Reason: Calculated potential significantly underestimated the experimentally measured onset potential
Time resolved spectroscopic studies of iridium-based catalysts for water electrolysis · Imperial · PhD
Tried and failed
multiconfiguration pair-density functional theory applied to alkene torsional barrier estimation. Outcome: worse than baseline. Reason: standard on-top functionals with minimal active space significantly overestimate torsional barrier heights
Reconstructing dynamic correlation and bonding in curved pi-systems · Iowa State · PhD
Tried and failed
density functional theory state splitting models applied to interfacial electron transfer rate prediction. Reason: Predicted minor slowdown without accounting for steric and molecular tilting restrictions causing large experimental rate reductions
Designing nanocrystal donor : molecular acceptor interfaces for improved charge and energy transfer · UT Austin · PhD
Self-consistent field iterations and structural optimizations suffer from numerical instability and convergence failure
Small orbital energy gaps cause orbital occupation oscillations, while unconstrained excited states experience orbital reordering during self-consistent field iterations. Molecular dynamics on thin heterostructures, transition state searches, and relaxation of generated crystal frameworks also fail to achieve numerical convergence or energy equilibrium.
Tried and failed
density functional theory molecular dynamics applied to ultrathin heterostructure junction interfaces. Outcome: did not converge. Reason: numerical convergence was unstable during simulation of the thin multilayer structure
New materials for magnetic tunnel junction devices for neuromorphic and probabilistic computing · UT Austin · PhD
Tried and failed
direct excited-state density functional theory applied to defect electronic states in crystals. Outcome: did not converge. Reason: orbitals reorder during self-consistent field iterations without orbital sorting or spin-orbit coupling
Quantum Engineering with Mechanical Strain · Harvard · PhD
Tried and failed
density functional theory transition state optimization applied to chemical reaction saddle point search. Outcome: did not converge. Reason: Saddle points could not be located using the specific functional and basis set combination.
Developing predictive tools for solvent effects on thermodynamics and kinetics · MIT · PhD
Tried and failed
generative crystal structure modeling with density functional relaxation applied to hypothetical metal-organic framework structures. Outcome: did not converge. Reason: generated structures failed to reach energy equilibrium due to likely unrealistic predicted topology
Advances in Artificial Intelligence for Accelerated Discovery of Energy Storage Polymers · Georgia Tech · PhD
Tried and failed
density functional theory excited state calculation applied to nearly degenerate defect electronic states. Outcome: did not converge. Reason: small orbital energy gaps caused orbital occupation oscillations during self-consistent field iterations
Towards Engineering Point Defects for Quantum Information · Harvard · PhD
Quantum mechanical calculations face prohibitive computational costs when scaling to large molecular and soft-matter systems
Density functional theory is computationally intractable for modeling full device structures, large aqueous soft-matter systems, and full conformational libraries. Long timescale simulations and free energy landscape explorations with explicit solvent are abandoned due to these extreme computational demands.
Considered and rejected
Considered and rejected: Rejected Density Functional Theory (DFT) as the primary computational simulation technique for aqueous soft-matter systems due to excessive initial conditions and computational intractability.
STUDYING THE INTERFACIAL STRUCTURE OF HYDRATED SOFT MATTER TEMPLATES VIA FLUID CELL ATOMIC FORCE MICROSCOPY (AFM) TECHNIQUES · Cornell · PhD
Considered and rejected
Considered and rejected: Rejected Density Functional Theory (DFT) in favor of the Effective Mass Approximation (EMA) because modeling entire device structures with DFT requires excessive computational power.
Study of the temperature dependent electron mobility in GaN/ScAlN heterostructures · Boston University · Master's
Considered and rejected
Considered and rejected: Rejected Quantum Mechanics / DFT for full conformational library screening due to excessive computational cost, opting for Molecular Mechanics
Considered and rejected
Considered and rejected: Rejected using quantum DFT calculations for full free energy landscape due to high computational cost and inability to model >1000 explicit water molecules over 100+ ns.
Computational Investigations of Surfactants at Interfaces: Molecular Modeling and Machine Learning-Driven Discovery · Virginia Tech · Doctorate
Simplified electronic references and truncation approximations degrade accuracy in multireference and nonadiabatic dynamics
Spin-restricted references fail to capture electron correlation in conjugated polymers, and density functional multireference diagnostics do not correlate with correlation recovery. In excited state and nonadiabatic modeling, omitting residual nonadiabatic couplings or applying simplified Casida approximations causes incorrect wavepacket populations and inconsistent optical absorption orderings.
Tried and failed
density functional theory multireference diagnostics applied to predicting correlation energy recovery in molecules. Outcome: no signal. Reason: poor linear correlation between DFT diagnostics and multireference correlation energy recovery in distorted organic molecules
Quantum Chemistry Meets Machine Learning: Autonomous Computational Workflow for Chemical Discovery · MIT · PhD
Tried and failed
spin-restricted quantum embedding with perturbation theory applied to conjugated polymer electronic structure. Outcome: worse than baseline. Reason: spin-restricted reference fails to capture electron correlation in conjugated systems compared to unrestricted references
Tried and failed
omitting residual nonadiabatic coupling terms applied to quasidiabatic molecular quantum dynamics. Outcome: worse than baseline. Reason: residual couplings outside strict diabatic basis significantly alter wavepacket populations and computed spectra
Tried and failed
simplified Casida approximations for optical transitions applied to coupled quantum defect complexes. Reason: approximations yielded inconsistent absorption intensity orderings across varying inter-defect separations
Engineering Quantum Optical Matter: Defects, Entanglement, and Chemical Reactivity · Harvard · PhD
Standard density functionals without dispersion corrections fail to describe non-covalent interactions and lattice energetics
Omitting dispersion corrections results in heavy underbinding of molecular crystal lattices and large systematic errors in induction-bound ion-pi binding energies. Automated relaxation algorithms also introduce significant lattice parameter errors when relying on static dispersion coefficient assumptions.
Tried and failed
density functional theory without dispersion corrections applied to molecular crystal lattice energy prediction. Outcome: worse than baseline. Reason: standard local/semilocal functionals fail to describe long-range van der Waals dispersion, heavily underbinding molecular crystals
Reliable and efficient parameter estimation methodologies for crystal structure prediction · Imperial · PhD
Tried and failed
Standard non-dispersion-corrected density functional approximations applied to induction-bound ion-pi systems. Outcome: did not generalise. Reason: Lack of dispersion and proper non-covalent interaction modeling causes large systematic errors in binding energies
Improving the Accuracy, Transferability, and Efficiency of Hybrid Density Functional Theory · Cornell · PhD
Considered and rejected
Considered and rejected: Automated variable-cell relaxation algorithms in Quantum ESPRESSO, rejected because static dispersion coefficient assumptions in XDM cause significant errors for inorganic lattice parameters.
Improved Charge Transfer and Barrier Lowering across a Au-MoS2 Interface through the Insertion of a Layered Ca2N Electride · Dalhousie · Master's
Semi-empirical treatments and functional constraint enforcement introduce substantial errors in molecular properties
Density functional tight binding and standalone semi-empirical methods yield large energy errors that misrank conformational stabilities and miscalculate thermochemistry. Enforcing theoretical lower bound constraints on exchange enhancement factors also degrades the accuracy of molecular formation energy predictions.
Tried and failed
density functional tight binding calculation applied to adsorbate conformational energy ranking. Reason: semi-empirical approximations yielded large energy errors and failed to correctly rank relative geometric stabilities
Machine-learning models for analysis of biomass reactions and prediction of reaction energies · Georgia Tech · PhD
Tried and failed
enforcing theoretical lower bound constraints on functional applied to exchange enhancement factor in density functionals. Outcome: worse than baseline. Reason: constraining the exchange factor to physically motivated limits degraded molecular formation energy prediction accuracy
Analyzing and improving electronic structure calculations of catalytic interfaces using density functional theory and machine learning · Georgia Tech · PhD
Tried and failed
semiempirical quantum chemistry methods alone applied to molecular thermochemistry prediction. Reason: inherent electronic structure approximations cause unacceptably high systematic errors without higher-level single-point energy corrections
Accelerating Autonomous Molecular Discovery through Automated Quantum Chemistry and Artificial Intelligence · MIT · PhD
Left open by the authors
Problems the authors named and did not get to.
Left open
Incorporate finite-temperature image charge forces into DFT-based quantum mechanical injection barrier calculations for metal-polymer interfaces. Blocker: Lacks specific mathematical formulation or methodology for combining finite-temperature dynamics with image charge corrections in DFT barriers.
First-principles simulation of charge injection at metal/polymer interfaces · Imperial · PhD
Left open
Perform a quantum analysis of extended nonlinear topological photonic systems by combining topological laser models with dynamical mean-field theory (DMFT). Blocker: Lacks specific target models, precise boundary conditions, or concrete implementation details for coupling topological lasing with OpenBDMFT.
Topology and Nonlinearity in Driven-Dissipative Photonic Lattices: Semiclassical and Quantum Approaches · SISSA · Doctorate
Left open
Reparameterize classical force field functional forms using modern quantum mechanical reference data or machine learning for crystal structure generation. Blocker: Lack of specific functional forms, target molecules, training datasets, or concrete parameterization methodology.
Assessment and Improvement of Structure Generation in Crystal Structure Prediction Protocols · Dalhousie · Master's
Left open
Develop fault-tolerant quantum computation schemes tailored to the leading error types in trapped-ion quantum computers. Blocker: Broad research direction without specified error models, target fault-tolerant protocols, or concrete performance benchmarks
Improving Circuit Performance in a Trapped-Ion Quantum Computer · Duke · PhD
Left open
Test higher-order and state-component-mixing interacting Hamiltonians in quantum algorithms for solving nonlinear differential equations. Blocker: Lack of specific Hamiltonian formulations or benchmark target equations in the description.
Quantum algorithms for image classification and nonlinear differential equations · Scuola Normale Superiore · PhD
Left open
Investigate quantum technological applications of highly mixed state entanglement and quantum discord in driven-dissipative Dicke models. Blocker: The objective is broad and lacks specific concrete application targets, metrics, or methodology.
Phase Transitions in Dipole-Dipole Interacting Atomic Systems · MIT · PhD
Left open
Compare the performance of different quantum master equations in predicting fluctuations and thermodynamic uncertainty relations. Blocker: None
Dissipative quantum systems: theoretical foundations and applications · Scuola Normale Superiore · PhD
Left open
Simulate the effects of non-solvable strong defect perturbations on the Kitaev quantum spin liquid state at varying defect densities. Blocker: The unfinished work lacks a specific model formulation, parameter set, or target numerical method beyond broad research directions
Emergence of Chirality from Crystalline Defects in Kitaev Spin Liquids · Georgia Tech · PhD
Left open
Benchmark the bilevel Bayesian optimization method on real expensive simulations like density functional theory or molecular dynamics. Blocker: None
Exploiting known structure in data-driven models of dynamical systems · UT Austin · PhD
Who has worked on this
Authors of the public theses cited on this page, most-cited first.
- Sushree Jagriti Sahoo · Georgia Tech · PhD · 2025
Analyzing and improving electronic structure calculations of catalytic interfaces using density functional theory and machine learning - Yiyuan Wang · Imperial · PhD · 2021
First-principles simulation of charge injection at metal/polymer interfaces - Fatemeh Borhani · Georgia Tech · PhD · 2026
Emergence of Chirality from Crystalline Defects in Kitaev Spin Liquids - Haoyang (Oscar) Wu · MIT · PhD · 2026
Accelerating Autonomous Molecular Discovery through Automated Quantum Chemistry and Artificial Intelligence - Xin Wang · Virginia Tech · Doctorate · 2026
Computational Investigations of Surfactants at Interfaces: Molecular Modeling and Machine Learning-Driven Discovery - Paolo Alessandro Xavier TOGNINI · Scuola Normale Superiore · PhD · 2026
Quantum algorithms for image classification and nonlinear differential equations - Ezra A. Alexander · MIT · PhD · 2025
Colloidal Semiconductor Nanocrystals: Tools For and Insights From First Principles Investigations - Colin Bundschu · Cornell · PhD · 2025
The Oxygen Reduction Reaction on Spinel Oxides in Alkaline Media - Katherine Nicole Ferreras · Iowa State · PhD · 2025
Reconstructing dynamic correlation and bonding in curved pi-systems - Bo-Shiang Yang · Boston University · Master's · 2025
Study of the temperature dependent electron mobility in GaN/ScAlN heterostructures - Caiwu Liang · Imperial · PhD · 2024
Time resolved spectroscopic studies of iridium-based catalysts for water electrolysis - Konrad Hedderick · Cornell · PhD · 2023
STUDYING THE INTERFACIAL STRUCTURE OF HYDRATED SOFT MATTER TEMPLATES VIA FLUID CELL ATOMIC FORCE MICROSCOPY (AFM) TECHNIQUES
Checking a claim in this area?
We can run the same search on any method or claim. If nothing turns up, we will say so, and that proves nothing on its own.