Text from the interactive explanation
Deciphering the infrared spectrum of the hydrated proton using full-dimensional quantum dynamics
David Mendive-Tapia, Christoph Schran, Banshi Das, Fabien Gatti, Markus Schröder, Dominik Marx, Oriol Vendrell
Original paperNature Chemistry · 2026 · The hydrated proton
Full quantum simulations of a proton held by six water molecules show how its infrared spectrum slides by more than a thousand wavenumbers when a single neighbouring water is pulled away — a stroboscopic movie of frozen frames.
Computed infrared response of the central proton
Computed frame · Rconst = 2.6 Å
Potential felt by the central proton
Spectra and potential: computed in the paper. Molecule: schematic (distances from SI Table S3). Glow: 1D sketch of the proton’s position on the computed potential. Between the seven computed distances the readout holds the nearest computed frame.
1 · The puzzle
Two textbook pictures, one smeared spectrum
Acids are acidic because they release excess protons into water. Textbooks draw each proton in one of two limiting ways: the Zundel cation, H5O2+, with the proton shared midway between two waters, or the Eigen cation, H9O4+, in which a hydronium ion (H3O+, a water that has taken the proton) is hydrogen-bonded to three more waters.
Cold, isolated clusters of each kind have sharp infrared spectra. The horizontal axis is the wavenumber in cm⁻¹: how many waves fit in a centimetre — higher means a faster vibration.
Measured · Johnson et al. 2014
Measured · Wolke et al. 2016; Esser et al. 2018
Measured · Heine et al. 2013 (grey: ill-resolved)
Liquid acid looks nothing like these. Measured difference spectra of protons in water show a broad, gently modulated “continuum” absorbing from about 1,000 to 3,500 cm⁻¹. In the paper’s words, it “cannot be readily explained by a simple superposition” of the cluster spectra. Try it.
Try to mix the liquid band
Your weighted sum of the three measured cluster spectra (chartreuse), over four measured liquid spectra.
All curves on this bench are measured. Cluster spectra resampled to 4 cm⁻¹ and averaged with your weights; liquid difference spectra as digitized in the authors’ dataset (references in the legend). Intensities are relative.
The missing ingredient is asymmetry . Around a real proton in water the neighbouring molecules are never perfectly balanced, and they keep rearranging. That is precisely what the stroboscope explores: hold one water at a chosen distance and compute the spectrum of that frozen, lopsided frame. Back to the stroboscope ↑
2 · The check
Theory meets experiment
Before pulling anything, the authors computed the spectrum of the undisturbed H+(H2O)6 cluster and compared it with the measured one. The computed curve follows the measurement across the whole range — with no adjustable parameters and no frequency shift applied; band positions agree within the simulation’s resolution of about 15 cm⁻¹.
Pick a band to see which motion produces it.
Motion animations are exaggerated illustrations of the assigned motion. Assignments from the paper’s Fig. 3b and Supplementary Table S2.
3 · A missing fingerprint
Why the Zundel’s signature doublet vanishes
The bare Zundel cation has a famous fingerprint: its proton band near 1,000 cm⁻¹ is split into two peaks. This is a Fermi resonance — two vibrations with nearly the same frequency mix and share their intensity, so one band becomes two. In bare Zundel, the proton’s back-and-forth motion along the O–O axis happens to match an overtone of the slow wagging of the two waters combined with the O–O stretch.
In H+(H2O)6 the doublet is gone. The four outer waters hinder the inner waters’ wagging. In bare Zundel that wagging is slow (about 100 cm⁻¹, with the overtone that couples to the proton at 378 cm⁻¹); in the extended cluster it appears just above 800 cm⁻¹ — a shift of roughly 400 cm⁻¹ that detunes the resonance. Pull all four outer waters away symmetrically and watch the doublet come back.
Schematic: the Zundel core with its four outer waters at the chosen distance. Spectra: computed response of the central proton (paper’s Fig. 4a).
What couples to what
From its computed potential energy surface, the paper works out how strongly the key coordinates of the cluster are correlated. The three inner-core motions behind the Fermi resonance (r, z, γ) couple strongly to the outer O–O distances (R). As the authors put it, the four solvating waters “hinder the dynamics of the inner Zundel core, especially the wagging motion (γ)”.
inner core blockouter-water blockinner–outer coupling
Correlation matrix re-plotted from the paper’s Extended Data Fig. 1b (computed from the potential energy surface).
4 · Back to the beaker
Reading the liquid’s smear
The authors lay their computed frames over the measured spectra of acid in water. Each coloured window of the liquid band lines up with a different kind of lopsided environment around the proton. Choose a window.
Liquid curves: measured difference spectra of aqueous H+ (Fig. 2d). Overlays: computed cluster frames (Fig. 4b,c), each scaled to fit.
Long exposure
Expose every computed frame — central proton and O–H stretches, all pulling distances — onto one plate.
Frames exposed: 0
Illustration of the idea in the paper’s Fig. 1a — an overlay of computed cluster frames, not a fitted liquid spectrum.
5 · Under the hood
How the movie was computed
51 coupled degrees of freedom. Every vibration of H+(H2O)6 is treated quantum mechanically, fully coupled and anharmonic — no normal-mode shortcuts.
A compressed wavefunction. The multilayer MCTDH method stores the 51-dimensional wavefunction as a layered tensor network, which makes an otherwise impossible quantum calculation affordable without discarding couplings.
Coupled-cluster-quality forces. Neural-network potential-energy and dipole surfaces trained on CCSD(T)-grade data, with test errors of 0.06 kJ mol⁻¹ per atom and 0.03 D.
Spectrum from motion. The infrared spectrum is the Fourier transform of the dipole–dipole correlation function over 2 ps, giving a resolution of about 15 cm⁻¹.
The pull. In each stroboscopic frame one outer water’s O–O distance Rconst is held fixed; the remaining 50 dimensions evolve quantum mechanically.
Does the neural network match coupled cluster?
Line: neural-network potential. Dots: coupled-cluster reference. Energy cuts along z, each relative to its own minimum (paper’s Extended Data Fig. 2).
Next on the authors’ list: the equally broad infrared continuum of hydroxide in water — alkaline solutions.
Sources
Sources & credits
David Mendive-Tapia, Christoph Schran, Banshi Das, Fabien Gatti, Markus Schröder, Dominik Marx & Oriol Vendrell. “Deciphering the infrared spectrum of the hydrated proton using full-dimensional quantum dynamics.” Nature Chemistry 18, 1773–1780 (2026). doi:10.1038/s41557-026-02209-3 · article page
Open access under CC BY 4.0.
Computed spectra, potential scans and correlation data re-plotted from the authors’ dataset (doi:10.5281/zenodo.20411426, CC BY 4.0); resampled to 4 cm⁻¹; colours and layout changed.
Measured spectrum of bare Zundel: Johnson et al., J. Chem. Phys. 2014 (paper ref. 58).
Measured spectrum of bare Eigen: Wolke et al., Science 2016; Esser et al., J. Phys. Chem. Lett. 2018 (refs 59, 60).
Measured spectrum of H+(H2O)6: Heine et al., J. Am. Chem. Soc. 2013 (ref. 51).
Measured liquid difference spectra: Fournier et al. 2018, Dahms et al. 2017, Daly et al. 2017, Thämer et al. 2015 (refs 47, 46, 34, 44).
Computed bare-Eigen reference spectrum: Schröder et al. 2022 (ref. 56).
Paper figure: Fig. 1a, Mendive-Tapia et al., Nat. Chem. 2026, CC BY 4.0.
The 3D and 2D molecules, the glowing proton cloud and the motion animations are illustrations: the cluster geometry is schematic (O–O distances from Supplementary Table S3), the cloud is a one-dimensional sketch computed on the paper’s potential cuts rather than the full 51-dimensional wavefunction, and vibrations are exaggerated.
Topics in this iPaper
Chemistry
the concept of pH in acid–base chemistry
Introduction, page 1 (p. 1773)
Physics
full-dimensional quantum dynamics simulations of the extended Zundel
Abstract, page 1 (p. 1773)
infrared spectroscopy
experimental infrared (IR) spectra of the solvated proton
Abstract, page 1 (p. 1773)
quantum dynamics
full-dimensional quantum dynamics simulations of the extended Zundel
Abstract, page 1 (p. 1773)
multiconfiguration time dependent hartree
multilayer multiconfiguration time-dependent Hartree (ML-MCTDH) method
Methods, 'Quantum molecular dynamics simulations', page 9
proton transfer
allows us to tackle proton transfer dynamics, including quantum effects
Methods, 'Quantum molecular dynamics simulations', page 9
neural network potential
neural network-based CCSD(T)-quality potential energy surface
Fig. 4 caption, page 5 (p. 1777)
solvation
provide a direct map between the asymmetric solvation environment
Abstract, page 1 (p. 1773)