Beyond the Structure: Quantum Chemistry on Neurosnap

A structure tells you what is connected. Quantum chemistry helps you ask what that molecule might do.

A familiar scaffold can still surprise you. Move a functional group and the charge distribution shifts. Change the solvent and a different environment becomes energetically favorable. Rotate a bond and the geometry you have been looking at may no longer be the lowest-energy candidate.

For scientists, these are useful clues. They help turn “this compound behaves differently” into a more specific question—and, ideally, a better next experiment.

We are expanding that view on Neurosnap with four new PySCF-powered services and a substantial addition to Conformer Generator: molecular electrostatics and orbitals, solvent effects, QM conformer refinement, UV–visible absorption, and vibrational thermochemistry. The initial collection focuses on small, neutral molecules, with settings and interpretation limits attached to each calculation.

Think of it as five new lenses for the same scientific curiosity.

Figure 1. Choose the question first. These capabilities address complementary molecular properties; the diagram is a guide to the collection, rather than an automatic calculation sequence.

Where are the electrons—and why should I care?

A molecular drawing gives every bond a tidy line. The electrons are less tidy, and their distribution is often where the interesting chemistry begins.

Molecular Electrostatics & Orbitals brings that distribution into view through electron density, signed electrostatic-potential fields, and frontier orbitals. Dipole moments and atom-level Mulliken charges add numerical context.

For medicinal chemists, these views can help frame questions about polar interactions and changes between analogues. For formulation and materials researchers, they offer another way to investigate how molecular features might influence interactions with the surrounding environment. A useful comparison starts with consistent geometries, methods, and settings.

Water makes a good first illustration. In the actual calculation below, the potential varies around the molecule, and its computed dipole magnitude is approximately 1.90 Debye. Seeing a familiar molecule this way makes the central idea immediate: one structure contains regions with quite different electrostatic character.

Figure 2. An original visualization of a saved service calculation: gas-phase water, PBE0/def2-SVP, evaluated at uploaded coordinates with explicit hydrogens. This is the grid slice nearest the molecular plane, with the color scale clipped at ±0.08 hartree/e for readability. It is a potential-field slice, not potential painted onto an electron-density surface or a binding prediction.

The orbitals add another perspective. HOMO and LUMO shapes can help you explore the spatial character of frontier electrons, while their energies provide method-dependent descriptors. Orbital phases are arbitrary, Mulliken charges depend on the basis set, and a HOMO–LUMO gap should not be read as a measured absorption energy.

What changes when the solvent changes?

A molecule does not experience “solvent” as a label on a bottle. It experiences an environment that changes its energetics.

Solvation & Solvent Effects uses the SMD continuum-solvation approach available in PySCF to compare a molecule across selected pure solvents. The same molecular geometry is used throughout the comparison, helping isolate the modeled solvent contribution. A chosen reference solvent makes the differences easier to interpret.

In the water example in Figure 3, the estimated SMD energy is about 1.69 kJ/mol higher in ethanol and 11.44 kJ/mol higher in acetone than in water, under the shared-geometry assumptions. Those differences describe this particular calculation. They are not a general solvent ranking for other molecules.

For a solvent-selection study, this gives you a tractable molecular question to investigate before broadening the experimental screen. Crystalline-drug solubility also depends on the solid, ionization, and other effects; the calculation does not include those contributions or establish a partition coefficient.

If that distinction sounds familiar, it connects directly to our formulation tool collection and COSMOPharm workflow. The tools provide complementary evidence for different parts of the formulation problem.

Give molecular shape another look

Most flexible molecules have more than one plausible geometry. An electronic property calculated for one conformer is a statement about that conformer, and sometimes the geometry deserves as much attention as the property itself.

Conformer Generator now includes QM Refinement. Generate a small candidate ensemble or upload conformers of the same molecule, refine them with DFT, and compare their relative electronic energies in gas phase or a selected continuum solvent.

This is useful when you want to examine whether an interpretation depends on a particular molecular shape. It can also help you select candidate geometries for closer study. The starting ensemble still matters: refining a handful of candidates cannot reveal conformers that were never sampled.

The rankings are electronic-energy comparisons. They do not include the full thermal and entropic contributions needed to establish solution populations, and the available functionals have no added empirical dispersion correction. Treat close rankings as an invitation to check sensitivity, especially where weak interactions matter.

Follow the light—and the motion

UV–Visible Absorption Prediction addresses the connection between electronic structure and light. It predicts vertical singlet excitations using TDA or TDDFT, reports wavelengths and oscillator strengths, and provides an illustrative broadened spectrum. Leading natural-transition-orbital pairs help visualize the hole and particle character associated with a transition. The PySCF excited-state documentation describes the underlying methods.

For chromophore comparisons and analytical research, that combination is useful: the transition list tells you where absorption is predicted, while the orbitals help you explore its character. The formaldehyde example below also shows why a transition energy alone is incomplete—one calculated transition has near-zero oscillator strength. Predicted absorption does not establish fluorescence yield, emission, or photostability.

Vibrational Frequencies & Thermochemistry turns to molecular motion. After gas-phase geometry optimization, it calculates harmonic modes, provides mode animations, and checks for imaginary frequencies. Ideal-gas rigid-rotor/harmonic-oscillator estimates add zero-point energy and thermal corrections at the selected temperature and pressure.

Water's three calculated vibrational modes provide a compact demonstration. The frequencies in Figure 3 are unscaled harmonic predictions; the equal-height markers locate modes and do not represent IR intensities. For larger flexible molecules, low-frequency motions can make harmonic entropy estimates especially sensitive. These molecular calculations do not supply crystal melting points or fusion enthalpies.

Figure 3. Saved calculation outputs, not experimental measurements or an accuracy benchmark. Left: water SMD energy differences at a common gas-optimized geometry. Center: three gas-phase formaldehyde TDA singlet transitions; marker height represents oscillator strength. Right: water's gas-phase harmonic frequencies, shown without intensities. All three examples use PBE0/def2-SVP; the water geometry in the solvent and vibration examples was optimized separately from the uploaded geometry in Figure 2. The absorption panel shows discrete transitions rather than the service's broadened display curve.

From a molecular input to an interpretable result

The common approach is straightforward: start with an appropriate molecular structure, prepare or supply its geometry, choose consistent electronic-structure settings, and calculate the property that answers your question. Where geometry optimization is selected, DFT refines the starting structure before the property calculation. Fixed-coordinate modes let you investigate a supplied explicit-hydrogen 3D geometry directly; vibrational analysis always optimizes first.

The collection offers PBE0 and B3LYP with def2-SVP or def2-TZVP. A larger basis provides a useful sensitivity check, with additional computational cost. Continuum solvent treatments describe an averaged environment; individual solvent molecules, specific complexes, and conformer ensembles require further investigation.

The initial scope is deliberately bounded: neutral, closed-shell molecules containing H, C, N, O, F, P, S, or Cl, without counterions or isotope labels. Most workflows allow up to 24 heavy atoms and 80 atoms in total; vibrational thermochemistry has a smaller limit of 12 heavy atoms and 40 total atoms. These boundaries help keep the scientific assumptions clear.

What have we checked?

Before presenting these capabilities, we ran real molecular calculations through each workflow and checked that the results retained their scientific meaning. Across the development validation, 38 focused checks passed, including 16 actual molecular job executions.

Those checks covered water electrostatics and charge balance, finite density and orbital fields, solvent-reference differences, conformer refinement, vibrational-mode outputs, and positive excitation energies with consistent wavelength conversion. We also exercised alternative calculation settings and checked that unsupported chemistry and invalid geometries are rejected. Result-page checks used captured calculation outputs to verify tables, plots, file selection, and behavior when a visualization cannot load.

This establishes useful evidence for calculation and result-handling behavior. Experimental predictive accuracy across broad chemical space remains a separate question. The examples here are small-molecule demonstrations, and the best way to build confidence for your own system is to compare settings, examine conformer sensitivity, and test against relevant measurements.

Bring a sharper question to the next experiment

A charge distribution can suggest what to investigate about an interaction. A solvent comparison can help focus a screening panel. A refined geometry can reveal how much a conclusion depends on shape. An absorption prediction or a vibrational mode can give a measurement a more specific molecular interpretation.

That is the appeal of this collection: more ways to connect a structure with a scientific hypothesis you can act on.

Explore the new quantum chemistry tools in the Neurosnap service catalog, or revisit our formulation overview to see how these molecular questions sit alongside solubility, ionization, and mixture thermodynamics.

Scientific credit: these workflows build on PySCF, molecular preparation with RDKit, and established electronic-structure, continuum-solvation, and harmonic-analysis methods. Figures are original Neurosnap illustrations; Figures 2 and 3 use actual saved calculation outputs. No experimental benchmark data are presented.