Sotiris S. Xantheas

The many-body expansions of Hartree–Fock (HF) and correlation contributions to water cluster MP2 binding energies converge differently: the HF contribution (one-body (1B) repulsive, two-body (2B) and three-body (3B) attractive) converges at the 3B term, whereas the MP2 correlation contribution (1B and 2B attractive) converges at the 2B term, with 99.7% arising from localized double excitations spanning one and two monomers. Two orbital excitation truncation approximations, MP2-1B2B (geometry based) and localized LMP2-1M2M (excitation space based), recover ≥99.2% of the canonical MP2 binding energies.

This study explores how the vibrational properties of a water molecule change as it becomes part of increasingly larger hydrogen-bonded networks of increasing solvation shells (first, second, and third), simulating environments mimicking liquid water. Using advanced ab initio methods [MP2 and CCSD(T)] and comparing them with classical interaction potentials, the analysis focuses on both the harmonic and anharmonic frequencies of the bending and stretching vibrations. The H–O–H bending mode quickly shifts to higher frequencies (blue shift) with solvation, rising from 1596 cm−1 in the monomer to 1653–1664 cm−1 in the first solvation shell, and stays within the observed infrared range for liquid water as more shells are added. In contrast, the O–H stretching frequencies experience significant red shifts, which depend strongly on the local hydrogen bond structure. Second-shell tetrahedral networks show large red shifts due to cooperative hydrogen bonding, while third-shell networks produce shifts typical of bulk water. The bending vibration is mainly influenced by immediate (nearest neighbor) hydrogen bond geometry and stabilizes quickly in tetrahedral environments, whereas the stretching modes reflect the broader distribution of hydrogen bond strengths that emerge in larger (beyond nearest neighbor) solvation shells. Overall, these findings deliver a systematic microscopic picture of how water’s vibrational spectrum transitions from an isolated molecule to a bulk-like environment, highlighting the importance of including an environment-dependent 1-body term in classical many-body polarizable potentials for water.

Valence CASSCF (vCAS) calculations for (H₂O)₂, (HF)₂, and HF-H₂O dimers fail to capture donor bond lengthening and associated redshift of its stretching frequency due to hydrogen bonding. Analysis of the orbitals in the vCAS calculation reveals that one of the active weakly occupied dimer orbitals correlates electrons in the acceptor lone pairs, not the donor bond, which is left without left-right correlation. This has significant implications for subsequent treatments of dynamical correlation that are based on a vCAS reference wave function.

We constructed a benchmark dataset (H₂X100) of 100 CCSD(T)/Complete Basis Set (CBS) binding energies for the (H₂X)_n, X = S, Se, Te, n = 2–4 clusters, and additional CCSD(T)-quality structures and harmonic vibrational frequencies for 12 unique homodimer minima, containing a mix of the hydrogen and chalcogen bonding motifs. Using the new H₂X100 dataset, we assessed the performance of 17 density functionals across the rungs of Jacob’s ladder, with and without dispersion corrections, spanning the GGA, mGGA, GH-GGA, GH-mGGA, RSH-GGA, and RSH-mGGA groupings. The r²SCAN-D4, B97M-rV, and ωB97M-V functionals provide the best agreement with the CCSD(T) benchmarks and can be used in studies of larger ensembles of these systems, including simulations of their condensed phases.

We report cutting-edge performance results via mixed-precision spin-adapted ab initio density matrix renormalization group (DMRG) electronic structure calculations utilizing the Ozaki scheme for emulating FP64 arithmetic through the use of fixed-point compute resources. By approximating the underlying matrix and tensor algebra with operations on a modest number of fixed-point representatives (“slices”), we demonstrate on smaller benchmark systems and for the active compounds of the FeMoco and cytochrome P450 (CYP) enzymes with complete active space (CAS) sizes of up to 113 electrons in 76 orbitals [CAS(113, 76)] and 63 electrons in 58 orbitals [CAS(63, 58)], respectively, that milli-Hartree accuracy can be reached with mixed-precision arithmetic. We also show that, due to its variational nature, DMRG provides an ideal tool to benchmark accuracy domains, as well as the performance of new hardware developments and related numerical libraries. Detailed numerical error analysis and performance assessment are also presented for subcomponents of the DMRG algebra by systematically interpolating between double- and pseudo-half-precision. Our analysis represents the first quantum chemistry evaluation of FP64 emulation for correlated calculations capable of achieving even chemical accuracy and emulation based on fixed-point arithmetic, and it paves the way for the utilization of state-of-the-art Blackwell technology in tree-like tensor network state electronic structure calculations, opening new research directions in materials sciences and beyond.

We assess the competing preference for hydrogen and chalcogen bonding in the aqueous heterodimers of the chalcogen hydrides H₂O-H₂X (X = S, Se, Te). We employ the CCSD(T) method near the complete basis set limit and include an explicit account of anharmonic zero-point energy corrections to establish accurate binding energies. We identify three bonding motifs in the H₂O-H₂X dimers, namely, H₂X (i) donating a hydrogen bond, (ii) accepting a hydrogen bond, and (iii) donating a chalcogen bond, and observe the emergence of chalcogen bonds to water for H₂Se and H₂Te. The hydrogen bond donated by a water molecule is consistently more stable than the hydrogen bond donated by the H₂X (X = S, Se, Te) molecule, with the strength of the former decreasing by a smaller margin than the latter down the chalcogen series. Conversely, the chalcogen bond becomes stronger progressing from H₂Se, in which it is weaker, to H₂Te, in which it is more stable than the two other bonding motifs. As a result, the global minimum for H₂O-H₂Te is chalcogen bonded. The analysis of the bonding using quasi-atomic orbitals (QUAOs) offers insights into the nature of the chalcogen- and hydrogen-bond accepting bonding motifs. Specifically, the water molecule prefers to accept hydrogen bonds from its mixed s/p hybrid lone pair and chalcogen bonds from its primarily p-hybridized lone pair. In contrast, the heavier chalcogen hydrides prefer to accept hydrogen bonds from their primarily p-hybridized lone pairs. A remarkable linear correlation exists between the QUAO-derived atomic charges localized on the donor and acceptor heavy atoms and the respective binding energies of the heterodimers.

The replacement of uridine (U) with pseudouridine (ψ) in mRNA vaccines reduces immune inflammation as a result of the molecular recognition of mRNA by one of the Toll-like Receptors (TLR7). This process, which makes the practical application of mRNA vaccines possible, is modulated at the molecular level by the protons shared between the carboxylate group of amino acid residues in TLR7 and the deprotonation sites on U or ψ that determine the charged state in which these molecules exist under physiological conditions. In the gas phase, U is deprotonated at the N3 site, while ψ is deprotonated at the N1 site. This difference originates from the distinct chemical environments of the respective nucleobase moieties rather than the ribose sugar group and can provide the stepping stone toward a molecular level insight into the recognition process in the complex biological environment.

We report cutting-edge performance results via mixed-precision spin adapted ab initio Density Matrix Renormalization Group (DMRG) electronic structure calculations utilizing the Ozaki scheme for emulating FP64 arithmetic through the use of fixed-point compute resources. By approximating the underlying matrix and tensor algebra with operations on a modest number of fixed-point representatives (``slices''), we demonstrate on smaller benchmark systems and for the active compounds of the FeMoco and cytochrome P450 (CYP) enzymes with complete active space (CAS) sizes of up to 113 electrons in 76 orbitals [CAS(113, 76)] and 63 electrons in 58 orbitals [CAS(63, 58)], respectively, that the chemical accuracy can be reached with mixed-precision arithmetic. We also show that, due to its variational nature, DMRG provides an ideal tool to benchmark accuracy domains, as well as the performance of new hardware developments and related numerical libraries. Detailed numerical error analysis and performance assessment are also presented for subcomponents of the DMRG algebra by systematically interpolating between double- and pseudo-half-precision. Our analyis represents the first quantum chemistry evaluation of FP64 emulation for correlated calculations capable of achieving chemical accuracy and emulation based on fixed-point arithmetic, and it paves the way for the utilization of state-of-the-art Blackwell technology in tree-like tensor network state electronic structure calculations, opening new research directions in materials sciences and beyond.