A deep neural network model for packing density predictions and its application in the study of 1.5 million organic molecules.

The process of developing new compounds and materials is increasingly driven by computational modeling and simulation, which allow us to characterize candidates before pursuing them in the laboratory. One of the non-trivial properties of interest for organic materials is their packing in the bulk, which is highly dependent on their molecular structure. By controlling the latter, we can realize materials with a desired density (as well as other target properties). Molecular dynamics simulations are a popular and reasonably accurate way to compute the bulk density of molecules, however, since these calculations are computationally intensive, they are not a practically viable option for high-throughput screening studies that assess material candidates on a massive scale. In this work, we employ machine learning to develop a data-derived prediction model that is an alternative to physics-based simulations, and we utilize it for the hyperscreening of 1.5 million small organic molecules as well as to gain insights into the relationship between structural makeup and packing density. We also use this study to analyze the learning curve of the employed neural network approach and gain empirical data on the dependence of model performance and training data size, which will inform future investigations.

Benchmarking DFT approaches for the calculation of polarizability inputs for refractive index predictions in organic polymers.

In a previous study, we introduced a new computational protocol to accurately predict the index of refraction (RI) of organic polymers using a combination of first-principles and data modeling. This protocol is based on the Lorentz-Lorenz equation and involves the calculation of static polarizabilities and number densities of oligomer sequences, which are extrapolated to the polymer limit. We chose to compute the polarizabilities within the density functional theory (DFT) framework using the PBE0/def2-TZVP-D3 model chemistry. While this ad hoc choice proved remarkably successful, it is also relatively expensive from a computational perspective. It represents the bottleneck step in the overall RI modeling protocol, thus limiting its utility for virtual high-throughput screening studies, in which efficiency is essential. For polymers that exhibit late-onset extensivity, the employed linear extrapolation scheme can require demanding calculations on long-oligomer sequences, thus becoming another bottleneck. In the work presented here, we benchmark DFT model chemistries to identify approaches that optimize the balance between accuracy and efficiency for this application domain. We compare results for conjugated and non-conjugated polymers, augment our original extrapolation approach with a non-linear option, analyze how the polarizability errors propagate into the RI predictions, and offer guidance for method selection.

Roaming-like Mechanism for Dehydration of Diol Radicals.

Diol radicals (DRs) are important intermediates in biocatalysis, atmospheric chemistry, and biomass combustion. They are particularly generated from photolysis of halogenated diols and addition of hydroxyl radical to a double bond of unsaturated alcohols, such as lignols. The energized DRs further isomerize/decompose to form products, including water. Aqueous-phase dehydration in radiolytic and biomimetic systems typically occurs at low temperatures, with or without catalysis, whereas the gas-phase dehydration is usually considered energetically unfavorable. In the present study, we propose a new low-energy, roaming-like mechanism based on a detailed dispersion-corrected DFT and ab initio level analysis of the gas-phase dehydration of DRs obtained from the combination of OH radicals with allyl alcohol (AA, CH2═CHCH2OH)-the simplest relevant model of the unsaturated alcohols. The roaming pathways involve a nearly dissociated OH-group, which subsequently abstracts an H atom of the remaining fragment to form water and [C3H5O] radical via a transition state (TS) with energy close to the C-O bond fission asymptote. Two types of roaming-like first-order saddle points (SP) are identified for unimolecular dehydration of 1,2- and 1,3-DR radical adducts involving either both hydroxyl groups of diol radicals to generate an oxygen-centered radical, or ß-OH group and a skeletal α-hydrogen atom of the 1,2-DR to form a resonantly stabilized hydroxyallyl radical. Two higher energy conventional (tight) transition states, along with the pathways to 1,2-OH-migration, as well as direct H-abstraction, are also identified and analyzed. Most of the traditional density functional theory methods that have been successfully employed in the literature to locate so-far-known roaming SPs were also able to identify the new mechanism, in accord with dispersion-corrected double hybrid B2PLYP-D3(BJ) and mPW2PLYPD methods involving MP2-correlation corrections. However, the MP2 method itself failed to locate any of them, which seems to be typical for MP2 method for loose TS structures, confirmed here for a flat region of PES connecting direct and roaming saddle points. However, MP2 method correctly locates an identical roaming SP for a larger p-coumaryl alcohol model involving hydroxyphenyl substituent at Cγ atom of AA. Two types of interfragmental interactions are identified that stabilize the roaming SPs: (a) H-bonding of the leaving OH radical either with the H atom of the remaining OH group, or with π-cloud of the double bond; (b) direct interaction of π-electrons with the lone-pair electrons of the heteroatom in the leaving OH group through the TS-ring. The alternative TSs are qualitatively characterized by "collinearity" angle of the OH radical attack on the O-H/C-H bonds of the substrate in abstraction-like O-H-O geometry, attributed to the improved orbital overlaps. The proposed mechanism presents broader implications to signify, particularly, a larger role in atmospheric and combustion processes, especially biomass pyrolysis.

Combining first-principles and data modeling for the accurate prediction of the refractive index of organic polymers.

Organic materials with a high index of refraction (RI) are attracting considerable interest due to their potential application in optic and optoelectronic devices. However, most of these applications require an RI value of 1.7 or larger, while typical carbon-based polymers only exhibit values in the range of 1.3-1.5. This paper introduces an efficient computational protocol for the accurate prediction of RI values in polymers to facilitate in silico studies that can guide the discovery and design of next-generation high-RI materials. Our protocol is based on the Lorentz-Lorenz equation and is parametrized by the polarizability and number density values of a given candidate compound. In the proposed scheme, we compute the former using first-principles electronic structure theory and the latter using an approximation based on van der Waals volumes. The critical parameter in the number density approximation is the packing fraction of the bulk polymer, for which we have devised a machine learning model. We demonstrate the performance of the proposed RI protocol by testing its predictions against the experimentally known RI values of 112 optical polymers. Our approach to combine first-principles and data modeling emerges as both a successful and a highly economical path to determining the RI values for a wide range of organic polymers.

Revisiting the polytopal rearrangements in penta-coordinate d7-metallocomplexes: modified Berry pseudorotation, octahedral switch, and butterfly isomerization.

This paper provides a first-principles theoretical investigation of the polytopal rearrangements and fluxional behavior of five-coordinate d7-transition metal complexes. Our work is primarily based on a potential energy surface analysis of the iron tetracarbonyl hydride radical HFe˙(CO)4. We demonstrate the existence of distorted coordination geometries in this prototypical system and, for the first time, introduce three general rearrangement mechanisms, which account for the non-ideal coordination. The first of these mechanisms constitutes a modified version of the Berry pseudorotation via a square-based pyramidal C4v transition state that connects two chemically identical edge-bridged tetrahedral stereoisomers of C2v symmetry. It differs from the classical Berry mechanism, which involves two regular D3h equilibrium structures and a C4v transition state. The second mechanism is related to the famous "tetrahedral jump" hypothesis, postulated by Muetterties for a number of d6 HML4 and H2ML4 complexes. Here, our study suggests two fluxional rearrangement pathways via distinct types of C2v transition states. Both pathways of this mechanism can be described as a single-ligand migration to a vacant position of an "octahedron", thus interchanging (switching) the apical and basal ligands of the initial quasi-square pyramidal isomer, which is considered as an idealized octahedron with a vacancy. Accordingly, we call this mechanism "octahedral switch". The third mechanism follows a butterfly-type isomerization featuring a key-angle deformation, and we thus call it "butterfly isomerization". It connects the quasi-square pyramidal and edge-bridged tetrahedral isomers of HFe˙(CO)4 through a distorted edge-bridged tetrahedral transition state of Cs symmetry. Our paper discusses the overall features of the isomers and rearrangement mechanisms as well as their implications. We rationalize the existence of each stationary point through an electronic structure analysis and argue their relevance for isolobal analogues of HFe˙(CO)4.

The Harvard organic photovoltaic dataset.

The Harvard Organic Photovoltaic Dataset (HOPV15) presented in this work is a collation of experimental photovoltaic data from the literature, and corresponding quantum-chemical calculations performed over a range of conformers, each with quantum chemical results using a variety of density functionals and basis sets. It is anticipated that this dataset will be of use in both relating electronic structure calculations to experimental observations through the generation of calibration schemes, as well as for the creation of new semi-empirical methods and the benchmarking of current and future model chemistries for organic electronic applications.

A theoretical study of the 3d-M(smif)2 complexes: structure, magnetism, and oxidation states.

We carry out a theoretical investigation of the recently reported M(smif)(2) series1,2 and find a number of interesting phenomena. These include complex potential energy surfaces with near-degenerate stationary points, low-lying states, non-trivial electron configurations, as well as non-innocent ligand behavior. The M(smif)(2) exhibit a delicate balance between geometry and electronic structure, which has implications not only for their reactivity but also for controlling their properties through ligand design. We address methodological issues and show how conceptual quantities such as oxidation states and electronic configurations can be extracted through a simple analysis of the electron and spin densities-without a complicated examination of the underlying orbitals.

Analytic response theory for the density matrix renormalization group.

We propose an analytic response theory for the density matrix renormalization group, whereby response properties correspond to analytic derivatives of density matrix renormalization group observables with respect to the applied perturbations. Both static and frequency-dependent response theories are formulated and implemented. We evaluate our pilot implementation by calculating static and frequency-dependent polarizabilities of short oligodiacetylenes. The analytic response theory is competitive with dynamical density matrix renormalization group methods and yields significantly improved accuracies when using a small number of density matrix renormalization group states. Strengths and weaknesses of the analytic approach are discussed.

Orbital optimization in the density matrix renormalization group, with applications to polyenes and beta-carotene.

In previous work we have shown that the density matrix renormalization group (DMRG) enables near-exact calculations in active spaces much larger than are possible with traditional complete active space algorithms. Here, we implement orbital optimization with the DMRG to further allow the self-consistent improvement of the active orbitals, as is done in the complete active space self-consistent field (CASSCF) method. We use our resulting DMRG-CASSCF method to study the low-lying excited states of the all-trans polyenes up to C24H26 as well as beta-carotene, correlating with near-exact accuracy the optimized complete pi-valence space with up to 24 active electrons and orbitals, and analyze our results in the light of the recent discovery from resonance Raman experiments of new optically dark states in the spectrum.

The radical character of the acenes: a density matrix renormalization group study.

We present a detailed investigation of the acene series using high-level wave function theory. Our ab initio density matrix renormalization group algorithm has enabled us to carry out complete active space calculations on the acenes from napthalene to dodecacene correlating the full pi-valence space. While we find that the ground state is a singlet for all chain lengths, examination of several measures of radical character, including the natural orbitals, effective number of unpaired electrons, and various correlation functions, suggests that the longer acene ground states are polyradical in nature.

Targeted excited state algorithms.

To overcome the limitations of the traditional state-averaging approaches in excited state calculations, where one solves and represents all states between the ground state and excited state of interest, we have investigated a number of new excited state algorithms. Building on the work of van der Vorst and Sleijpen [SIAM J. Matrix Anal. Appl. 17, 401 (1996)], we have implemented harmonic Davidson and state-averaged harmonic Davidson algorithms within the context of the density matrix renormalization group (DMRG). We have assessed their accuracy and stability of convergence in complete-active-space DMRG calculations on the low-lying excited states in the acenes ranging from naphthalene to pentacene. We find that both algorithms offer increased accuracy over the traditional state-averaged Davidson approach, and, in particular, the state-averaged harmonic Davidson algorithm offers an optimal combination of accuracy and stability in convergence.

Multireference correlation in long molecules with the quadratic scaling density matrix renormalization group.

We have devised a local ab initio density matrix renormalization group algorithm to describe multireference correlations in large systems. For long molecules that are extended in one of their spatial dimensions, we can obtain an exact characterization of correlation, in the given basis, with a cost that scales only quadratically with the size of the system. The reduced scaling is achieved solely through integral screening and without the construction of correlation domains. We demonstrate the scaling, convergence, and robustness of the algorithm in polyenes and hydrogen chains. We converge to exact correlation energies (in the sense of full configuration interaction, with 1-10 microE(h) precision) in all cases and correlate up to 100 electrons in 100 active orbitals. We further use our algorithm to obtain exact energies for the metal-insulator transition in hydrogen chains and compare and contrast our results with those from conventional quantum chemical methods.