Liquid water has a dynamic atomic-scale structure, which gives rise to many of the unique properties of water, such as its extraordinarily high boiling and freezing points. Water’s hydrogen bonding—the interaction that attracts a hydrogen atom on one molecule to an oxygen atom on another—facilitates the transfer of a small amount of electric charge, about one-fiftieth that of a single electron, between molecules.
No two molecules have identical sets of hydrogen bonds, because each hydrogen bond affects the formation of others on the same molecule and beyond (see figure 1). The behavior yields a complex network of hydrogen bonds that are constantly forming and breaking on time scales of a millionth of a millionth of a second. Hydrogen bonds are complicated further by nuclear quantum effects (NQEs)—the position of a hydrogen atom, because of its low mass, is delocalized. Computations predict that NQEs can weaken hydrogen bonds.1
What is known about hydrogen bonds in liquid water comes predominantly from molecular dynamics simulations. Because the bonds carry a small amount of delocalized charge, which is transferred when the bonds are broken or formed, changes in pH could affect charge transfer in liquid water.2 Theorists have proposed that in a cluster of three water molecules, excess protons decrease or hydroxide ions increase the amount of electronic charge that is shifted across the hydrogen-bond network. NQEs should also affect charge transfer, but that possibility has not been well observed.
The lack of experimental data on charge transfer and NQEs is caused by the structural complexity of water and intrinsic limitations in the spectroscopic methods used to measure them. Now Mischa Flór and Sylvie Roke of EPFL in Switzerland and their colleagues have developed an experimental approach that directly measures hydrogen bonds’ stretch mode between interacting molecules.3 The new observations are the first of their kind and help disentangle how pH and NQEs contribute to charge transfer in water.
Hiding in plain sight
Hydrogen bonds have resonant frequencies at about 200 cm−1 (6 THz), a frequency that is challenging to probe with IR spectroscopy. Although Raman spectroscopy can make reliable measurements at that frequency, the spectrum is unstructured and notoriously difficult to interpret. “What are we seeing there?” says Roke. “If you can tell, you will be famous.”
In a 2022 paper, Roke and colleagues reported a clue for how to focus on hydrogen bonds. They were using a near-IR femtosecond laser pulse to study water’s structure, and the liquid target emitted light at the laser’s second harmonic.4 The researchers found that nonlinear spectroscopies, including frequency-doubling techniques, are sensitive probes of the transient, nonhomogeneous structure that hydrogen bonds provide liquid water over the duration of the probing laser pulse.
The remaining critical insight for how to experimentally isolate the hydrogen-bond signal came to Roke and Flór when they were discussing a 1966 theory paper that focused on nonlinear optical effects in homogeneous liquids.5 The paper derived the relationships among the second-harmonic emissions that are expected for the four possible polarization combinations of the ingoing and outgoing light.
Flór realized that if the relationships were applied to a liquid that has a transient structure arising from hydrogen bonding, they could be used to isolate the intensity from only the interacting molecules. “The equations have been there for 60 years, but there was just one missing small trick that had to be done,” says Flór. “And that’s what we found.” By measuring samples at each of the four polarization combinations, the researchers could simply obtain the signal from liquid water’s hydrogen bonds.
Nonlinear spectroscopy
With the analytical approach for second-harmonic scattering in place, Roke and colleagues developed a technique they call correlated vibrational spectroscopy. With good spectral resolution and at various scattering angles, it records spectra of scattered light at low frequencies.
The spectra contain peaks at the second-harmonic frequency minus the resonant frequency of a vibrational mode. The vibrational mode comes from a nonlinear optical process called hyper-Raman scattering. Paper coauthor David Wilkins, from Queen’s University Belfast in Northern Ireland, worked out the theoretical formalism to mathematically show that hyper-Raman scattering could be used to directly measure the frequency of the hydrogen-bond stretch mode.
From spectra measured for different polarization combinations, the researchers calculated one combination that contained only single-molecule interactions and one that contained only signatures of interacting molecules. As shown in figure 2, the hydrogen-bond stretch mode was visible only in the interacting-molecule spectrum, as expected.
The frequency of the hydrogen-bond stretch mode contains information about charge transfer. The bond strength, which is measured by the frequency, is linearly related to the bond’s electric charge. Any shifts in the frequency, then, yield information about charge transfer.
The group’s experiments on acidic and basic solutions of water found a difference in charge transfer. Compared with water with a pH of 7, highly acidic water contained 4% less electronic charge in the hydrogen-bond network, and extremely basic water contained 8% more electronic charge in the hydrogen-bond network. The results agree with quantum chemical computations of small water clusters.
In addition to considering pH, Flór, Roke, and colleagues also explored how NQEs affect the charge transfer in water’s hydrogen-bonding network. They analyzed sample solutions of heavy and normal water and found that the frequency of the hydrogen-bond stretch mode in heavy water was shifted higher than that of normal water. Hydrogen, being lighter than deuterium, is much more sensitive to NQEs and is more delocalized. Those features resulted in a 10% weaker bond.
Beyond liquids
For the method’s proof of concept, the researchers’ main focus was on water. But other liquids—and even 2D or 3D solids—could be studied too. In one of their experiments, the researchers studied the molecular vibrations of a solution of potassium thiocyanate salt in water. As expected, the vibrational modes of the molecular ions were observed only in the single-molecule spectrum, and changes to the hydrogen-bond network were seen in the interacting-molecule spectrum.
Correlated vibrational spectroscopy, the researchers say, could help improve the understanding of how water’s structure affects and promotes biochemical reactions associated with DNA and proteins. The researchers are analyzing the results from several other liquid samples to better understand the capabilities of the approach. “Every time we measured a sample, we got a whole truckload of information,” says Roke. “We are still working through publishing all the other things that we measured.”
References
- D. M. Wilkins et al., J. Phys. Chem. Lett. 8, 2602 (2017).
- M. Soniat, R. Kumar, S. W. Rick, J. Chem. Phys. 143, 044702 (2015).
- M. Flór et al., Science 386, eads4369 (2024).
- T. Schönfeldová et al., J. Phys. Chem. Lett. 13, 7462 (2022).
- R. Bersohn, Y.-H. Pao, H. L. Frisch, J. Chem. Phys. 45, 3184 (1966).
