arXiv subjects

Avetik R. Harutyunyan

Publications and source records attributed to Avetik R. Harutyunyan.

7 recordsLinked to original sources

arXiv2026

Coupled-cluster molecular properties across the main group that extrapolate beyond training size

Coupled-cluster theory defines the accuracy standard for molecular electronic-structure properties but scales too steeply for routine application, whereas density-functional theory is affordable yet systematically biased. We resolve this trade-off with a single equivariant network, HARP (Hamiltonian Read-out for Properties), that predicts an effective one-electron Hamiltonian from one inexpensive B3LYP/def2-SVP calculation and derives a broad suite of properties from it (energy, optical gap, dipole, quadrupole, polarizability, Mulliken atomic charges, and Mayer bond orders) at coupled-cluster accuracy across nine main-group elements, including the under-served phosphorus, sulfur, and chlorine chemistries. The model is trained on a new in-house dataset of multi-property labels computed at the CCSD(T) level for all nine elements. On a held-out test set, it reduces the error of every property by a factor of 3.8 to 270 relative to semi-local, hybrid, and double-hybrid DFT (referenced to composite CCSD(T)/cc-pVTZ), while adding only ~0.1 s wall time per molecule, delivering coupled-cluster-quality predictions at the cost of a single DFT calculation. Critically, deriving every property from a predicted Hamiltonian rather than pooling per-atom features builds the correct size-scaling into the model architecture: on pi-conjugated oligothiophenes it matches finite-field CCSD polarizability to ~1% and the EOM-CCSD optical gap to ~3% at the largest sizes where those references remain affordable (44 and 37 atoms, where a single CCSD field point already costs ~500x the model's entire inference) and extrapolates the corrected trends to 58-atom chains, a regime where pooling-based architectures fail by construction. Accurate extrapolation is therefore set by the model's inductive bias rather than by the training data.

physics.chem-ph↗

arXiv2024

Multi-task learning for molecular electronic structure approaching coupled-cluster accuracy

Machine learning (ML) plays an important role in quantum chemistry, providing fast-to-evaluate predictive models for various properties of molecules. However, most existing ML models for molecular electronic properties use density functional theory (DFT) databases as ground truth in training, and their prediction accuracy cannot surpass that of DFT. In this work, we developed a unified ML method for electronic structures of organic molecules using the gold-standard CCSD(T) calculations as training data. Tested on hydrocarbon molecules, our model outperforms DFT with the widely-used hybrid and double hybrid functionals in computational costs and prediction accuracy of various quantum chemical properties. As case studies, we apply the model to aromatic compounds and semiconducting polymers on both ground state and excited state properties, demonstrating its accuracy and generalization capability to complex systems that are hard to calculate using CCSD(T)-level methods.

physics.chem-ph↗

arXiv2014

Insights into carbon nanotube nucleation: Cap formation governed by catalyst interfacial step flow

In order to accommodate an increasing demand for carbon nanotubes (CNTs) with desirable characteristics one has to understand the origin of helicity of their structures. Here, through in situ microscopy we demonstrate that the nucleation of a carbon nanotube is initiated by the formation of the carbon cap. Nucleation begins with the formation of a graphene embryo that is bound between opposite step-edges on the nickel catalyst surface. The embryo grows larger as the step-edges migrate along the surface, leading to the formation of a curved carbon cap when the steps flow across the edges of adjacent facets. Further motion of the steps away from the catalyst tip with attached rims of the carbon cap generates the wall of the nanotube. Density Functional Theory calculations bring further insight into the process, showing that step flow occurs by surface self diffusion of the nickel atoms via a step-edge attachment-detachment mechanism. Since the fact that cap forms first in the sequence of stages involved in nanotube growth, we suggest that it originates the helicity of the nanotube. Therefore, the angular distribution of catalyst facets could be exploited as a new parameter for controlling the curvature of the cap and, presumably, the helicity of the nanotube.

cond-mat.mtrl-sci↗

arXiv2014

Probing the Electrical Properties of Overlapped Graphene Grain Boundaries by Raman spectroscopy

The effect of grain boundaries and wrinkles on the electrical properties of polycrystalline graphene is pronounced. Here we investigate the stitching between grains of polycrystalline graphene, specifically, overlapping of layers at the boundaries, grown by chemical vapor deposition (CVD) and subsequently doped by the oxidized Cu substrate. We analyze overlapped regions between 60 and 220 nm wide via Raman spectroscopy, and find that some of these overlapped boundaries contain AB stacked bilayers. The Raman spectra from the overlapped grain boundaries are distinctly different from bilayer graphene and exhibit splitting of the G band peak. The degree of splitting, peak widths, as well as peak intensities depend on the width of the overlap. We attribute these features to inhomogeneous doping by charge carriers (holes) across the overlapped regions via the oxidized Cu substrate. As a result, the Fermi level at the overlapped grain boundaries lies between 0.3 and 0.4 eV below the charge neutrality point. Our results suggest an enhancement of electrical conductivity across overlapped grain boundaries, similar to previously observed measurements(1). The dependence of charge distribution on the width of overlapping of grain boundaries may have strong implications for the growth of large-area graphene with enhanced conductivity.

cond-mat.mtrl-sci↗

arXiv2013

Feasibility of Lithium Storage on Graphene and Its Derivatives

Nanomaterials are anticipated to be promising storage media, owing to their high surface-to-mass ratio. The high hydrogen capacity achieved by using graphene has reinforced this opinion and motivated investigations of the possibility to use it to store another important energy carrier - lithium (Li). While the first-principles computations show that the Li capacity of pristine graphene, limited by Li clustering and phase separation, is lower than that offered by Li intercalation in graphite, we explore the feasibility of modifying graphene for better Li storage. It is found that certain structural defects in graphene can bind Li stably, yet more efficacious approach is through substitution doping with boron (B). In particular, the layered C3B compound stands out as a promising Li storage medium. The monolayer C3B has a capacity of 714 mAh/g (as Li1.25C3B), and the capacity of stacked C3B is 857 mAh/g (as Li1.5C3B), which is about twice as large as graphite's 372 mAh/g (as LiC6). Our results help clarify the mechanism of Li storage in low-dimensional materials, and shed light on the rational design of nano-architectures for energy storage.

cond-mat.mtrl-sci↗

arXiv2008

Influence of Mo on the Fe:Mo:C nano-catalyst thermodynamics for single-walled carbon nanotube growth

We explore the role of Mo in Fe:Mo nanocatalyst thermodynamics for low-temperature chemical vapor deposition growth of single walled carbon nanotubes (SWCNTs). By using the size-pressure approximation and ab initio modeling, we prove that for both Fe-rich (~80% Fe or more) and Mo-rich (~50% Mo or more) Fe:Mo clusters, the presence of carbon in the cluster causes nucleation of Mo2C. This enhances the activity of the particle since it releases Fe, which is initially bound in a stable Fe:Mo phase, so that it can catalyze SWCNT growth. Furthermore, the presence of small concentrations of Mo reduce the lower size limit of low-temperature steady-state growth from ~0.58nm for pure Fe particles to ~0.52nm. Our ab initio-thermodynamic modeling explains experimental results and establishes a new direction to search for better catalysts.

cond-mat.mtrl-sci↗

arXiv2006

Theoretical study of the thermal behavior of free and alumina-supported Fe-C nanoparticles

The thermal behavior of free and alumina-supported iron-carbon nanoparticles is investigated via molecular dynamics simulations, in which the effect of the substrate is treated with a simple Morse potential fitted to ab initio data. We observe that the presence of the substrate raises the melting temperature of medium and large $Fe_{1-x}C_x$ nanoparticles ($x$ = 0-0.16, $N$ = 80-1000, non- magic numbers) by 40-60 K; it also plays an important role in defining the ground state of smaller Fe nanoparticles ($N$ = 50-80). The main focus of our study is the investigation of Fe-C phase diagrams as a function of the nanoparticle size. We find that as the cluster size decreases in the 1.1-1.6-nm-diameter range the eutectic point shifts significantly not only toward lower temperatures, as expected from the Gibbs-Thomson law, but also toward lower concentrations of C. The strong dependence of the maximum C solubility on the Fe-C cluster size may have important implications for the catalytic growth of carbon nanotubes by chemical vapor deposition.

cond-mat.mtrl-sci↗