Water – the molecule of life – is an intriguing substance to study. The properties of water, already unique at the macroscale, become even more perplexing when confined to nanoscale regimes. At these reduced length scales, water can exhibit anomalously low dielectric constants, support superionic phase behaviour, and display atypical friction responses, among other surprising phenomena. A deeper understanding of water’s nanoconfined behaviour could soon unlock a myriad of promising applications, including blue energy harvesting and next-generation desalination projects.

In the Michaelides ICE group, substantial effort has been directed towards understanding the properties of water under nanoconfinement [1,2]. Our research leverages machine-learning-enhanced molecular dynamics (MD) simulations, enabling us to attain ab initio-level predictions over large spatial and temporal scales – important for modelling realistic systems!

Recent work in our group has utilized the se methods towards understanding the behaviour of acidic and basic water in the vicinity of graphene-water interfaces [3]. Employing multi-nanosecond MD simulations, our findings reveal that protons accumulate at the graphene−water interface, with the hydronium ion predominantly residing in the first contact layer of water. In contrast, the hydroxide ion exhibits a bimodal distribution, found both near the surface and further away from it. Analysis of the underlying electronic structure reveals local polarization effects, resulting in counterintuitive charge rearrangement.

Beyond acidic water, we have also investigated the effect of nanoconfinement on electrolytic solutions. In one study, we have shown that the free energy of ion pairing substantially deviates from that in bulk solution, observing a decrease in contact ion pairing but an increase in solvent-separated ion pairing [4]. We attribute this to an interplay of ion solvation effects and graphene’s electronic structure. In our most recent work, we have shown ionic conductivity decreases as the degree of confinement increases, a trend governed by changes in both ion self-diffusion and dynamic ion–ion correlations [5]. This is accompanied by a shift in the ions’ diffusion mechanism toward more vehicular motion as the degree of confinement increases.

The bottleneck for all this work comes with the expensive ab initio calculations required to generate our models’ training data. As system size and complexity grow, so too do the costs of the requisite first-principles calculations. ARCHER2 has been instrumental in overcoming this barrier. With access to these HPC resources, we have been able to utilize larger and more representative datasets, improving model accuracy and mitigating finite-size effects. This has proved crucial for our nanoconfined work, in which we typically require sizeable boxes to support complex nanoconfined systems.

Several interesting questions remain in this compelling field. One such question is how to best utilize nanoconfinement for the adsorption of atmospheric gases. In collaboration with experimental spectroscopists, we are currently modelling the uptake of atmospheric CO₂ using nanoporous carbon electrodes. Our current results suggest an ‘oversolubility’ effect for CO₂, driven by an enhanced adsorption at the pore walls compared to water. Parallel to this work, we are exploring the influence of defects in nanoporous environments. Intrinsic defects in the form of single and double vacancies serve as hydrophilic centres in an otherwise hydrophobic graphene plane; a greater understanding of these effects may help facilitate the design of functionalized surface to promote desired reactions or adsorption processes.

Our work has helped identify several fundamental questions: How do reactivity, phase separation, and transport properties change under differing nanoconfinements? How do certain solutes adsorb at the graphene-water interface, and can we use this to facilitate mass transport? What is the influence of defects on these systems? How soon before we can actualize new technologies based on aqueous nanoconfinement? Utilizing ARCHER2 resources, we soon hope to answer many of these questions, helping drive the development of next-generation technologies across a range of applications and fields.

For additional information on our nanoconfined simulations and other research interests, visit: https://www.ch.cam.ac.uk/group/michaelides/index

Outcomes

• New knowledge of the behaviour of acidic and basic water under confinement.
• Determination of ion-pairing tendencies for nanoconfined electrolyte solutions.
• Estimates of the ionic conductivity and its relationship with confinement width.
• Established protocols for data generation and model training.
• Collaborative computational-experimental work targeting CO₂ adsorption in nanoporous environments.

Contact

Prof. Angelos Michaelides: am452@cam.ac.uk

References

1. Venkat Kapil, Christoph Schran, Andrea Zen, Ji Chen, Chris J. Pickard and Angelos Michaelides, Nature, 609, 512–516 (2022)
2. Fabian L. Thiemann, Christoph Schran, Patrick Rowe, Erich A. Müller and Angelos Michaelides, ACS Nano, 16, 7, 10775–10782 (2022).
3. Xavier R. Advincula, Kara D. Fong, Angelos Michaelides and Christoph Schran, ACS Nano (2025).
4. Kara D. Fong, Barbara Sumić, Niamh O’Neill, Christoph Schran, Clare P. Grey, and Angelos Michaelides, Nano Lett., 24, 16, 5024–5030 (2024).
5. Kara D. Fong, Clare P. Grey and Angelos Michaelides, ACS Nano, 19, 13, 13191–13201 (2025).