225. Property-tuned, defective, and amorphous MOF adsorbents via ligand tethering. D. Sensharma, V.V. Singh,

        S. Pal, S.C.B. Chackerian, F. Paesani, S.M. Cohen. ACS Appl. Mater. Interfaces 18, 31486 (2026). [link]

The covalent tethering of individual ligands into oligomers prior to metal–organic framework (MOF) synthesis can result in

unique structural effects in the resulting oligoMOFs. In this study, oligoMOFs based on the Cu-DMOF-1 framework are

synthesized using a series of flexible alkyl-tethered ligand dimers ranging in length from propyl to heptyl. Despite synthesis

under identical conditions, these Cu-oligoDMOFs show systematic variations in crystallinity, crystal size, thermal properties,

and porosity as a function of tether length. Through microcrystal electron diffraction (uED), powder X-ray diffraction (PXRD),

imaging, spectroscopy, and thermal analysis, in combination with periodic DFT calculations, we find that the balance and

interplay between the energetics of the metal ion coordination environment and the ligand constraints dictate the synthetic

outcomes of these self-assembly reactions. Increasingly severe geometric constraints imposed by shorter tethers result in

defective and even amorphous Cu-oligoDMOFs directly from synthesis, thereby outlining a novel, systematic, and reticular

approach by which MOF properties, including crystallinity, decomposition temperature, and defectivity, can be incrementally

tailored solely through tether sizing.

227. Microsolvation of protonated glycine: Infrared spectra from data-driven quantum many-body simulations.

        Z.A. Solomon, R. Rashmi, R. Zhou, F. Paesani. Under review. [link]

Understanding how hydration reshapes the structure and conformational flexibility of biomolecular ions is essential for

connecting gas-phase spectroscopy to behavior in aqueous environments. Here, we develop a data-driven many-body

potential energy function for GlyH+–H2O interactions and combine it with replica-exchange molecular dynamics to identify

isomeric equilibria and with temperature-elevated path-integral coarse-graining to include nuclear quantum effects in

simulations of GlyH+(H2O)n clusters. This framework captures many-body interactions with high-level ab initio accuracy and

enables direct computation of infrared spectra for comparison with experiment. By applying an inverse spectral

reconstruction of isomeric ensembles, we quantitatively decompose the experimental spectra into contributions from

competing hydration motifs and extract their relative populations. Our results characterize the sequential formation of the

first and second solvation shells, quantify the competition between intramolecular and water-mediated hydrogen bonds,

and reveal temperature dependence and nuclear quantum effects. Overall, this study provides a transferable approach to

understanding the hydration of biomolecular systems from gas-phase clusters to bulk aqueous solutions.

219. Surface-specific and bulk field-induced contributions to vibrational sum-frequency generation spectra at

        charged graphene oxide/water interfaces. T. O Balogun, R. Rashmi, G. Azom, R. Kumar, F. Paesani.

        J. Phys. Chem. Lett. 17, 2464 (2026). [link]

In this study, we use molecular dynamics simulations with the many-body MB-pol water potential to examine GO/water

interfaces with tunable surface charge. Nuclear quantum effects are incorporated using temperature-elevated path-integral

coarse-graining simulations enabling direct computation of both homodyne-detected (intensity) and heterodyne-detected

vibrational sum-frequency generation (vSFG) spectra. By explicitly accounting for the bulk third-order susceptibility, χ(3),

we show that the prominent lower-frequency band near 3200 cm−1 in experimental vSFG spectra is dominated by the

field-induced bulk χ(3) response, whereas the surfacespecific features are governed by the second-order susceptibility,

χ(2). These results clarify how surface-specific and bulk field-induced responses combine to shape vSFG spectra at

GO/water interfaces, providing a molecular basis for assigning spectral features to distinct interfacial hydrogen-bonding

motifs.

221. QUICK and robust ESP and RESP charges for computational biochemistry: Open-source GPU

        implementation. V. Tripathy, E. Palos, K.M. Merz, F. Paesani, A.W. Götz, J. Chem. Inf. Model. 66, 3173 (2026). [link]

We describe the implementation details of highly efficient ab initio electrostatic potential (ESP) calculations on graphics

processing units (GPUs), and introduce a novel scheme for partial charges that are robust against molecular orientation.

Performance analyses are discussed, and we highlight that in our new implementation, a single data center GPU can

outperform 128 corresponding data center CPU cores in time to solution. This implementation in the open-source

Quantum Interaction Computational Kernel code (QUICK) enables ESP computations on highly dense grids that surpass

what is reported in the literature, on the order of N(grid points) ~20000 points/atom. We demonstrate that, in this

dense-grid limit, ESP charges become independent of molecular orientation. We denote such ESP charges as being

robust against molecular orientation and validate this desirable attribute against standard charge schemes. Our proposed

charge scheme, called reweighted RESP (rwRESP), is designed to significantly overcome the sensitivity to N(grid points)

that limits the reliability of canonical RESP charges. By effectively amending this N(grid points)-sensitivity, we demonstrate

that rwRESP charges also achieve robustness against molecular orientation.

226. Many-body interactions govern halide distribution at the air/water interface.

        H. Agnew, S. Dasgupta, F. Paesani. Under review. [link]

Ion-specific surface propensities at the air/water interface remain debated. Here we use data-driven many-body energy

(MB-nrg) potentials to compute Helmholtz free-energy profiles for the halides using a controlled model hierarchy.

A polarizable baseline (TTM-nrg) reproduces the strong interfacial stabilization of I– (and smaller minima for Br– and Cl–)

commonly found with empirical polarizable force fields. In contrast, the fully many-body MB-nrg treatment, which adds

explicit 2-body and 3-body ion--water terms to many-body polarization, strongly suppresses interfacial adsorption across

the series: only I– retains a shallow minimum on the order of kBT near the Gibbs dividing surface, where kB is Boltzmann’s

constant. Decomposition of Helmholtz free-energy shows that this residual preference arises from a narrow energetic

minimum in the water--water contribution at the interface that compensates ion--water desolvation, while an opposing

entropic penalty compresses the net well. These results reconcile classical image-charge exclusion with polarizable-anion

continuum and interfacial-fluctuation models, and identify the collective water response, captured only with a quantitative

many-body description, as the mechanism that governs halide surface propensity.

© Paesani Research Group. All rights reserved.

220. Heterogeneity of free O-H groups at the air/water interface. R. Rashmi, K.-Y. Chiang, Y. Nagata, M. Bonn,

        F. Paesani, J. Phys. Chem. Lett. 17, 3013 (2026). [link]

Free O-H groups at the air/water interface produce a narrow peak in vibrational sum-frequency generation (vSFG) spectra,

yet their orientational anisotropy is often overlooked. Using polarization-resolved vSFG experiments and path-integral quantum

dynamics with the MB-pol potential, we show that the free O-H frequency depends on orientation: O-H bonds tilted toward the

interfacial plane are red-shifted and reorient faster than more upright O-H. This shift originates from orientation-dependent local

electric fields generated by neighboring water molecules, as demonstrated by a clear anti-correlation between the projected field

along the O--H bond and the fundamental frequency. Combining polarization-resolved vSFG with quantum data-driven many-body

simulations enables direct spectral interpretation in terms of orientation-dependent local electric fields, yielding a molecular-level

picture of interfacial free O--H heterogeneity.

218. Carbon polarization and carbon nanotube stacking impacts the behavior of water in nanoconfinement.

        L. Dick, R. Rashmi, H. Agnew, S. Saha, X. Zhu, B. Kirchner, F. Paesani, ACS Nano 20, 1304 (2026). [link]

The structural and dynamical properties of water confined in sub-nanometer, single-walled carbon nanotubes (CNTs) are

investigated using molecular dynamics simulations. With radii ranging from 4 to 8 Å, we assess confinement-size effects

using fully flexible CNTs with explicit carbon polarizability and the MB-pol water model. Density distributions reveal that

single-file water chains form in the narrow (10,0) CNT, pentagonal ring structures form in the intermediate (15,0) CNT, and

a two-layer arrangement forms in the wider (20,0) CNT. Angular distributions show that the influence of carbon polarization

on water orientation grows with CNT diameter, accompanied by a pronounced tilt of water molecules toward the CNT wall.

Comparisons with the TIP4P/2005f model indicate that while this classical model captures structural trends, MB-pol affords

a deeper description of interfacial water behavior by enabling direct simulations of infrared spectra from dipole trajectories

and consistently accounting for polarization effects. Finally, simulations of vertically aligned CNT arrays demonstrate that

inter-tube interactions significantly impact water dynamics: in stacked (20,0) configurations, intermittent hydrogen-bond

lifetimes decrease by 20-25% and layer residence times drop by up to 20% relative to isolated CNTs.

224. Hydration free energies of alkali metal and halide ions from data-driven many-body potentials.

        S. Saha, F. Paesani. J. Chem. Theory Comput. 22, 4688 (2026). [link]

Single-ion hydration free energies stringently test molecular models for aqueous ions, but quantitative comparison is

complicated by the interplay of short-range ion--water interactions, long-range electrostatics, many-body polarization,

and nuclear quantum effects (NQE). Here, we compute hydration free energies for alkali-metal cations (Li+–Cs+) and

halide anions (F––I–) using MB-nrg ion potentials in MB-pol water. Free energies are evaluated with a staged alchemical

cycle in which ion–water interactions are introduced sequentially (charge, polarization, and explicit 2-body and 3-body

terms), enabling stable sampling of each contribution. To perform robust charging transformations within the MBX

electrostatics framework, we implement a soft-core Coulomb scaling and evaluate free-energy changes using finite-

difference thermodynamic integration. Across both ion series, MB-nrg/MB-pol reproduces the expected monotonic

weakening of hydration with increasing ionic size and yields close agreement with experiment. Ion–water radial

distribution functions show that electrostatics rapidly establishes the first hydration shell, while the explicit many-body

corrections relax short-range overstructuring toward the fully interacting reference distribution.

222. Catching water’s hidden transition. F. Paesani, Science 391, 1318 (2026). [link]

Water behaves in unusual ways when it is cooled below its normal freezing point (0 °C) without turning into ice,

a phenomenon known as supercooling. In this regime, its density decreases after passing through a maximum,

the amount of energy required to raise its temperature (heat capacity) increases sharply, and water becomes

markedly more compressible. These anomalies point to an underlying reorganization of the hydrogen-bond network.

Probing these structural transformations has long been challenging because deeply supercooled water (i,e., liquid

water cooled well below its freezing point without crystallizing) solidifies on microsecond timescales, far too quickly

for conventional experimental techniques. In this Perspective, we discuss a novel experimental approach that

captures the structure of water in a transient liquid state just before freezing. These measurements reveal structural

and thermodynamic signatures consistent with our MB-pol predictions, significantly narrowing the range over which

a hidden liquid–liquid transition in supercooled water may occur.

223. From potentials to properties: Data-driven many-body simulations of water and aqueous systems. F. Paesani,

        R. Rashmi, H. Agnew, X. Zhu, E.F. Bull-Vulpe, R. Zhou, Annu. Rev. Phys. Chem. 77, 321 (2026). [link]

Water is one of the most extensively studied molecular systems, yet its behavior across different phases, interfaces,

and chemical environments continues to challenge existing models. Over the past decade, the development of

data-driven many-body potential energy functions (PEFs) within the many-body energy (MB-nrg) formalism has

enabled simulations of water and aqueous systems with unprecedented predictive power. Rooted in the many-body

expansion and rigorously derived from “gold standard” electronic structure data, these PEFs bridge quantum

chemistry and statistical mechanics within a unified framework. In this article, we review how the MB-pol PEF for

water and the MB-nrg PEFs for hydrated halide and alkali metal ions have reshaped our understanding of aqueous

properties across the gas, liquid, and solid phases, offering detailed insights into hydrogen bonding, spectroscopy,

isotope effects, and phase stability. By connecting length scales and timescales while maintaining quantum-

mechanical accuracy, the data-driven MB-nrg formalism provides a robust foundation for realistic molecular

simulations and offers new opportunities for addressing long-standing questions in physical chemistry and beyond.

228. Computational redesign of an antifreeze protein using deep learning. C. Calia, A.J. Altunc, R.J. Eufemio,

        B.O. Alvarado, J.D. Huynh, E. Oh, M.D. Burkart, K. Meister, F. Paesani. Under review. [link]

Antifreeze proteins (AFPs) found in various cold-adapted organisms inhibit ice growth and are of interest for applications

in food products, cryopreservation, agriculture, and materials science. Although high-resolution structures are available

for several AFPs, the amino acids required for full antifreeze activity remain incompletely defined, and the development of

AFP variants with properties such as enhanced solubility, high expression yield, and improved thermostability may further

facilitate applications. Here, we used the deep learning model ProteinMPNN to redesign the globular fish antifreeze

protein AFPIII, keeping the previously reported ice-binding residues fixed. We readily obtained sequences confidently

predicted to adopt AFPIII’s structure and we selected five designed variants for expression, all of which expressed

efficiently in E. coli. Circular dichroism spectroscopy showed that two of these variants retained secondary structure

elements consistent with AFPIII, whereas the other three exhibited structural differences. One design was predicted and

experimentally confirmed to have increased thermostability. All five variants displayed measurable thermal hysteresis

activity, although none reached the activity of wild-type AFPIII.

229. Memory-aware SIMD-enabled permutationally invariant polynomial code-generation procedure: Acceleration

        of the MB-pol potential energy function as an example. S. Iyer, E.F. Bull-Vulpe, C. Knight, F. Paesani.

        Under review. [link]

Permutationally invariant polynomials have been used to develop a variety of highly accurate potential energy

surfaces. Particularly when combined with the many-body expansion, such potentials present a promising avenue

for reducing the computational cost of simulations with coupled-cluster accuracy. In practice, performing molecular

dynamics simulations with many-body polynomial potential energy functions requires evaluating the polynomial

and its gradients thousands of times at each simulation step. Each individual gradient calculation requires

evaluating tens of thousands of floating-point operations, making polynomial evaluation an excellent candidate for

SIMD vectorization available on modern CPUs. A new memory-aware procedure for generating highly optimized

polynomial routines is presented and applied to the MB-pol potential energy function as a prototypical example.

These memory optimizations allow vector instructions to be efficiently exploited, producing speedups in polynomial

evaluation of nearly 4× and 8× when the AVX2 and AVX-512 vector instruction sets are employed, respectively.