Software solutions to the challenges of materials modelling

Ultralong Organic Phosphorescence (UOP) has generated significant interest because of its applications in several areas such as photovoltaic cells, bioimaging and anticounterfeiting.¹ The molecule of carbazole (Cz) is commonly used as a building block in organic materials for optoelectronic applications, acting as a light-absorbing, electron donor and emitting moiety. Cz and its derivatives display UOP at room temperature.  While the processes behind UOP have been associated with the stabilisation of H-aggregates, recent experimental studies indicate the presence of impurities drives the mechanism keeping the excited states alive for a long time.²  In this talk, Dr Crespo Otero will discuss the mechanism behind light-induced processes in crystalline and impure Cz and some derivatives using embedding methods developed in our group.^3,4 Dr Crespo Otero will revisit the role of aggregation and isomeric impurities on the excited state pathways and analyse the mechanisms for exciton, Dexter energy transfer and electron transport considering Marcus and Marcus–Levich–Jortner theories.⁵ The researchers' excited state mechanisms provide a plausible interpretation of the experimental results and support the formation of charge-separated states at the defect/host interface. They believe these results contribute to a better understanding of the factors that enhance the excited-state lifetimes in organic materials and the role of doping with organic molecules. 
 

References

¹ Kenry, C. Chen and B. Liu, Nat. Commun., 2019, 10, 2111. 

² C. Chen, Z. Chi, K. C. Chong, A. S. Batsanov, Z. Yang, Z. Mao, Z. Yang and B. Liu, Nat. Mater., 2021, 20, 175–180. 

³ M. Rivera, M. Dommett, A. Sidat, W. Rahim and R. Crespo‐Otero, J. Comput. Chem., 2020, 41, 1045–1058. 

⁴ M. Rivera, M. Dommett and R. Crespo-Otero, J. Chem. Theory Comput., 2019, 15, 2504–2516. 

⁵ F. J. Hernández and R. Crespo-Otero, J. Mater. Chem. C, 2021, 9, 11882–11892. 

ChemShell is a scriptable computational chemistry environment with an emphasis on multiscale simulation of complex systems using combined quantum mechanical and molecular mechanical (QM/MM) methods. The use of QM/MM embedded cluster calculations for studying materials systems will be highlighted with recent results on copper-containing zeolites, where selective catalytic reduction with ammonia can reduce the emission of environmentally harmful nitrogen oxides. Insights from QM/MM calculations on the influence of solvent on the SCR mechanism will be discussed. The recent major redevelopment of ChemShell as an open source, python-based package will also be presented, including implementation of new QM/MM schemes for materials modelling and the use of multi-level parallelisation frameworks on high performance computing platforms.

Kinetic Monte-Carlo (KMC) simulations have been instrumental in multiscale catalysis studies, enabling the elucidation of the complex dynamics of heterogeneous catalysts and the prediction of macroscopic performance metrics, such as activity and selectivity. However, the accessible length- and time-scales are still limited, and handling lattices containing millions of sites with 'traditional' sequential KMC implementations becomes prohibitive due to large memory requirements and long simulation times. Domain decomposition approaches could address these limitations, but they are challenging to implement in KMC simulation due to the inherently sequential nature thereof, by which one reactive event is causally linked to future (and past) events. These causal relations and the random time advancements of KMC steps, necessitate sophisticated protocols for conflict resolution at the boundaries between subdomains. Jefferson’s Time-Warp algorithm overcomes these challenges using local operations: sending and receiving messages/anti-messages, saving simulation state snapshots, and rolling-back in time to reinstate a previous state. Thus, any causality violations, arising transiently during simulation, are corrected, and the exact dynamics of the underlying stochastic model (the master equation) are finally reproduced. In this work, the researchers have coupled the Time-Warp algorithm with the Graph-Theoretical KMC framework enabling the handling of complex adsorbate lateral interactions and reaction events within large lattices. This approach has been implemented in Zacros, their general-purpose KMC software, and has been validated and benchmarked for efficiency in model systems as well as realistic chemistries. This work makes Zacros the first-of-its-kind general-purpose KMC code with distributed parallelisation capability to study heterogeneous catalysts.

Modeling has two main objectives, the rationalisation of existing results and the prediction of new systems and new phenomena. The field has made tremendous progress, and the availability of powerful computers, sophisticated software solutions, and machine learning algorithms are contributing to the discovery of new materials. However, the widespread use of modeling tools can also lead to inaccurate or useless predictions with potentially negative effects on the credibility of the field. In this talk Professor Pacchioni will briefly illustrate this emerging problem using materials for catalysis as an example. A large number of variables determine the performance of a heterogeneous catalyst, posing a challenge for quantum chemical modeling. The complexity of the problem has been significantly reduced with the advent of single atom catalysts (SACs) and, in particular, graphene-based SACs. In this context we are witnessing an increasing tendency to screen large number of materials based on density functional theory (DFT) and to propose universal descriptors to provide a guide for the synthesis of new catalysts. Professor Pacchioni will critically analyse some of the current problems associated with this activity: accuracy of calculations, neglect of important contributions in the models used, physical meaning of the proposed descriptors, imprecise data sets used to train machine learning algorithms, not to mention a few serious data reproducibility problems. The final message is that in an era where computing power has grown enormously, perhaps the time has come to shift the focus from the quantity to the quality of the data produced, to be of real help to the experimentalist in the design of new catalytic ¹

¹G. Di Liberto, Luis A. Cipriano, G. Pacchioni, “Universal principles for the rational design of single atom electrocatalysts? Handle with care”, ACS Catalysis, 12, 5846-5856 (2022).

Metals alloys are known to improve the performance of heterogenous catalysts for many different reactions, such as CO₂ hydrogenation^1,2. Predicting an effective and novel catalyst, by computing combinations of elements and compositions using Density Functional Theory (DFT), is expensive and highly time-consuming. To accelerate the discovery process, one can couple DFT with cluster expansion (CE), which the authors demonstrate here in the identification of an effective and novel catalyst for CO₂ hydrogenation. The authors focus on the PdZn (101) and (110) facets, which are most stable for binary alloys, and CE models for both of facets were tested in the concentration range where the body-centred tetragonal (BCT) phase exists, which is from 40 to 50% Zn concentration. These results show that higher Pd concentrations are predicted for both facets in the sub-surface while the top layers remain chemically ordered in a 1:1 alloy; the behaviour changes outside the stable BCT range, where segregation of Zn on the top surface is predicted for Zn concentrations above 50%. The outcomes highlight the critical importance of CE approach for determining structure−property relationships and accelerating the discovery and design of surfaces and nanostructured materials.

¹GA Olah, Angew. Chem., Int. Ed. (2005), 44, 2636.
²F Brix et al, The Journal of Physical Chemistry Letters 2020 11 (18), 7672-7678

Ice is one of the most interesting molecular crystals for a wide range of scientific fields, including physics, chemistry, biology, as well as space and material science. Its electronic properties are determined by complex topologies of hydrogen bonds as well as non-local dispersion interactions, providing a stern set to test electronic structure theory methods and get general insights both on molecular crystals and liquid water, where high accuracy simulations are hindered by prohibitive computational cost. Moreover, the recent discoveries of new ice phases have renewed the interest and increased the already extreme complexity of the water phase diagram, calling for a new analysis on the performance of electronic structure methods on the stability of the known polymorphs. The DMC-ICE13 dataset was constructed by computing reference lattice energies of thirteen ice polymorphs with sub-chemical accuracy using Diffusion Monte Carlo and conducting an extensive benchmark of several Density Functional Theory functionals. The obtained results suggest that a single functional achieving reliable performances for all phases is still missing, and that care is needed in the selection of the most appropriate functional for the desired application.

Dissolution is a ubiquitous phenomenon, and important biological and physical processes rely on dissolution of ionic salts in water. Simulations lend themselves conveniently to studying dissolution, providing the spatio-temporal resolution that can be difficult or impossible to obtain experimentally. Nevertheless, it's a complex process, requiring ab initio level of treatment and long time simulations, typically hindered by computational expense. The authors use advances in machine learning potential methodology to resolve the entire dissolution mechanism of NaCl in water under multiple conditions with an ab initio equivalent accuracy. A general overall mechanism pervades whereby a steady ion-wise unwrapping of the crystal precedes its rapid disintegration. This disintegration is characterised by the crystal reaching a point of maximum instability, where destabilising ion-water interactions dominate over the cohesive ion-ion interactions of the crystal. From high to low concentrations, dissolution times can be categorised as highly stochastic to almost completely deterministic respectively. They show the stochastic extent of the overall mechanism can be reasoned by simple arguments with respect to microscopic sub-events during the dissolution. This study provides important insight into a long-standing fundamental scientific problem. Moreover, this has important implications on our general understanding of dissolution in other crystals, as well as crystal nucleation.