Research

Our group uses theory, coarse-grained simulations, information geometry, and machine learning to study ionic liquids, polymers, and charged soft matter. We also apply machine learning to environmental data.

1 Theory of ion-containing polymers and liquids

We develop statistical-mechanical theories of ion-containing polymers and liquids, such as salt-doped polymers, ionic liquids, and polymerized ionic liquids (PILs). Our theories describe how ions, molecular dipoles, and chain connectivity affect phase behavior, dielectric response, and ion solvation.

Our earlier theories address the effects of dielectric inhomogeneity and electrostatic correlation on the solvation energy of ions (Nakamura, J. Phys. Chem. B 122, 6064–6071 (2018)), water dissolution in ionic liquids between charged surfaces (Chen et al., Mol. Syst. Des. Eng. 3, 328–341 (2018)), and microphase separation of ionic liquid-containing diblock copolymers (Nakamura, Macromolecules 53, 3891–3899 (2020)). We have also reviewed theoretical aspects of ionic liquids (Nakamura et al., Isr. J. Chem. 59, 813–823 (2019)) and the polarization of ionic liquids and polymers (Gao et al., J. Polym. Sci. 59, 2434–2457 (2021)).

We study the thermodynamic properties of polyelectrolytes, ionic liquids, and PILs, and develop coarse-grained theories of ion solvation in dielectrically inhomogeneous media. We are developing a Landau theory for PILs and comparing it with the polymeric Stockmayer fluid molecular dynamics simulations described below. We also use the Fisher–Rao geometry of the Flory–Huggins structure factor in the random-phase approximation to design the temperature schedule of scattering measurements near a spinodal, and we test the design on virtual experiments.

Two panels, each with a cation at the center, drawn as a large blue bead with a plus sign. Left: separate gray beads, each carrying an arrow, around the ion. Right: gray beads joined into chains, each bead carrying an arrow. In both panels the four arrows nearest the ion sit inside a pale disk and point away from it.
Schematic of ion solvation by molecular dipoles in a liquid (left) and in a polymer whose chain segments carry dipoles (right). The central bead with a plus sign is a cation, and arrows denote molecular dipoles. The dipoles nearest the ion are drawn aligned with its electric field.
Area 1 references
  1. Gao, T., Itliong, J. N., Kumar, S. P., Hjorth, Z., and Nakamura, I., “Polarization of ionic liquid and polymer and its implications for polymerized ionic liquids: An overview towards a new theory and simulation”, J. Polym. Sci. 59, 2434–2457 (2021).
  2. Nakamura, I., “Microphase Separation of Ionic Liquid-Containing Diblock Copolymers: Effects of Dielectric Inhomogeneity and Asymmetry in the Molecular Volumes and Interactions between the Cation and Anion”, Macromolecules 53, 3891–3899 (2020).
  3. Nakamura, I., “Phase diagrams of polymer-containing liquid mixtures with a theory-embedded neural network”, New J. Phys. 22, 015001 (2020).
  4. Shock, C. J., Stevens, M. J., Frischknecht, A. L., and Nakamura, I., “Solvation Energy of Ions in a Stockmayer Fluid”, J. Phys. Chem. B 124, 4598–4604 (2020).
  5. Nakamura, I., Shock, C. J., Eggart, L., and Gao, T., “Theoretical Aspects of Ionic Liquids for Soft-Matter Sciences”, Isr. J. Chem. 59, 813–823 (2019).
  6. Nakamura, I., “Effects of Dielectric Inhomogeneity and Electrostatic Correlation on the Solvation Energy of Ions in Liquids”, J. Phys. Chem. B 122, 6064–6071 (2018).
  7. Chen, H., An, L., and Nakamura, I., “Water dissolution in ionic liquids between charged surfaces: effects of electric polarization and electrostatic correlation”, Mol. Syst. Des. Eng. 3, 328–341 (2018).
  8. Liu, L. and Nakamura, I., “Solvation Energy of Ions in Polymers: Effects of Chain Length and Connectivity on Saturated Dipoles near Ions”, J. Phys. Chem. B 121, 3142–3150 (2017).
  9. Nakamura, I., “Spinodal Decomposition of a Polymer and Ionic Liquid Mixture: Effects of Electrostatic Interactions and Hydrogen Bonds on Phase Instability”, Macromolecules 49, 690–699 (2016).
  10. Ren, C.-L., Nakamura, I., and Wang, Z.-G., “Effects of Ion-Induced Cross-Linking on the Phase Behavior in Salt-Doped Polymer Blends”, Macromolecules 49, 425–431 (2016).
  11. Nakamura, I., “Dipolar Self-Consistent Field Theory for Ionic Liquids: Effects of Dielectric Inhomogeneity in Ionic Liquids between Charged Plates”, J. Phys. Chem. C 119, 7086–7094 (2015).
  12. Chen, H. and Nakamura, I., “Effects of the Dielectric Response of Single-Component Liquids and Liquid Mixtures on Electrochemical Properties between Charged Plates”, J. Phys. Chem. C 119, 24714–24723 (2015).
  13. Duan, X. and Nakamura, I., “A new lattice Monte Carlo simulation for dielectric saturation in ion-containing liquids”, Soft Matter 11, 3566–3571 (2015).
  14. Nakamura, I., “Ion Solvation in Polymer Blends and Block Copolymer Melts: Effects of Chain Length and Connectivity on the Reorganization of Dipoles”, J. Phys. Chem. B 118, 5787–5796 (2014).
  15. Nakamura, I., “Synergistic effects of ion pairs on the dielectric properties of diblock copolymer melts”, Soft Matter 10, 9596–9600 (2014).
  16. Nakamura, I. and Wang, Z.-G., “Thermodynamics of Salt-Doped Block Copolymers”, ACS Macro Lett. 3, 708–711 (2014).
  17. Thelen, J. L., Teran, A. A., Wang, X., Garetz, B. A., Nakamura, I., Wang, Z.-G., and Balsara, N. P., “Phase Behavior of a Block Copolymer/Salt Mixture through the Order-to-Disorder Transition”, Macromolecules 47, 2666–2673 (2014).
  18. Wang, X., Thelen, J. L., Teran, A. A., Chintapalli, M., Nakamura, I., Wang, Z.-G., Newstein, M. C., Balsara, N. P., and Garetz, B. A., “Evolution of Grain Structure during Disorder-to-Order Transitions in a Block Copolymer/Salt Mixture Studied by Depolarized Light Scattering”, Macromolecules 47, 5784–5792 (2014).
  19. Nakamura, I., Balsara, N. P., and Wang, Z.-G., “First-Order Disordered-to-Lamellar Phase Transition in Lithium Salt-Doped Block Copolymers”, ACS Macro Lett. 2, 478–481 (2013).
  20. Nakamura, I. and Wang, Z.-G., “Effects of dielectric inhomogeneity in polyelectrolyte solutions”, Soft Matter 9, 5686–5690 (2013).

2 Ionic liquids and polymerized ionic liquids

Solvent-free polymerized ionic liquids (PILs) combine molecular dipoles, mobile ions, and chain connectivity in a single condensed phase. We study ionic liquids and PILs with coarse-grained molecular dynamics simulations based on Stockmayer fluids.

With the Stockmayer fluid model, in which ionic groups are charged dipolar spheres, we compute the viscosity and glass transition temperature of ethylammonium nitrate (Itliong et al., J. Chem. Phys. 163, 044503 (2025)). We also compute the dielectric constants of ethylammonium nitrate and poly(N-vinyl ethylimidazolium bromide), modeling the latter with freely rotating and with restricted side-chain dipoles, and we relate the rotational freedom of the dipoles to the change in dielectric constant between a PIL and its monomeric ionic liquid (Shock et al., J. Phys. Chem. B 130, 1045–1053 (2026)). We examine how cation size and dipole mobility affect viscosity, chain dynamics, and ion pairing in PILs. We study correlated ion transport and structural reorganization in the poly(imidazolium) ionic liquid poly(EtVIM+)/TFSI− as the lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) loading increases, using species-resolved Onsager coefficients. Here, EtVIM+ denotes the N-vinyl ethylimidazolium cation and TFSI− the bis(trifluoromethanesulfonyl)imide anion. We plan to treat interfacial dielectric relaxation in PILs as a reorganization of polarization states, using an information-geometric theory.

Coarse-grained molecular dynamics of poly(EtVIM+)/TFSI− with LiTFSI at 350 K. Beads: backbone CH2 (blue) and CH (red), imidazolium (yellow), TFSI− (magenta), Li+ (green); short black lines mark dipoles.
Bead diagram: CH2 and CH backbone beads (σ2, σ1), an imidazolium bead (σ3) with charge q+ and dipole μ+, and an N(SO2CF3)2 bead (σ4) with charge q− and dipole μ−
Coarse-grained representation of a PIL monomer and its counterion, with bead diameters σ, charges q, and dipole moments μ.
Area 2 references
  1. Shock, C. J., Frischknecht, A. L., Stevens, M. J., and Nakamura, I., “Influence of Polymerization and Restricted Dipole Motion on the Dielectric Constants of Ionic Liquids”, J. Phys. Chem. B 130, 1045–1053 (2026).
  2. Itliong, J. N., Frischknecht, A. L., Stevens, M. J., and Nakamura, I., “Stockmayer fluid simulations for viscosity and glass transition temperature of ionic liquids”, J. Chem. Phys. 163, 044503 (2025).
  3. Shock, C. J., Stevens, M. J., Frischknecht, A. L., and Nakamura, I., “Molecular dynamics simulations of the dielectric constants of salt-free and salt-doped polar solvents”, J. Chem. Phys. 159, 134507 (2023).
  4. Gao, T., Itliong, J. N., Kumar, S. P., Hjorth, Z., and Nakamura, I., “Polarization of ionic liquid and polymer and its implications for polymerized ionic liquids: An overview towards a new theory and simulation”, J. Polym. Sci. 59, 2434–2457 (2021).
  5. Nakamura, I., “Microphase Separation of Ionic Liquid-Containing Diblock Copolymers: Effects of Dielectric Inhomogeneity and Asymmetry in the Molecular Volumes and Interactions between the Cation and Anion”, Macromolecules 53, 3891–3899 (2020).
  6. Shock, C. J., Stevens, M. J., Frischknecht, A. L., and Nakamura, I., “Solvation Energy of Ions in a Stockmayer Fluid”, J. Phys. Chem. B 124, 4598–4604 (2020).
  7. Liu, L. and Nakamura, I., “Solvation Energy of Ions in Polymers: Effects of Chain Length and Connectivity on Saturated Dipoles near Ions”, J. Phys. Chem. B 121, 3142–3150 (2017).

3 Information geometry of soft matter

We apply Fisher information and related information-geometric quantities to molecular simulation. A central quantity is the information-geometric speed (IGS), the Fisher–Rao speed of an evolving probability distribution.

We compute the IGS of time-dependent pair distributions from molecular dynamics simulations of Lennard-Jones systems at several thermodynamic states. For polymerized ionic liquids, we use the IGS, computed from coarse-grained molecular dynamics simulations of poly(EtVIM+)/TFSI− at different chain lengths, to study how ion hopping changes with chain length and relates to ionic conductivity. We also apply information-geometric quantities to phase separation in polymer blends.

Evolving state distribution Information- geometric speed Reorganization event 1 — select to move the marker here Reorganization event 2 — select to move the marker here reorganization events state s lag time t p(s, t) vIGS(t)

Schematic of an evolving state distribution p(s, t) (left) and the information-geometric speed vIGS(t) (right), whose peaks mark lag intervals of fastest distributional change.
Curved teal surface with a glowing path running across it
Conceptual illustration of a trajectory on a statistical manifold.
Cube with interpenetrating teal and white domains
Cube with teal droplets in a white matrix
connecteddispersed
Conceptual illustration of connected (left) and dispersed (right) domains in polymer-blend phase separation. Drag the divider to compare.

4 Machine learning

We construct surrogate models with artificial neural networks that reproduce the results of our coarse-grained molecular dynamics simulations (Gao et al., MRS Commun. 12, 966–974 (2022)). We use neural networks to learn statistical thermodynamic theories, such as the Landau theory of microphase separation for phase diagrams of diblock copolymers (Nakamura, New J. Phys. 22, 015001 (2020)). We apply machine learning to environmental data as well. Our recent study predicts oil contamination in water from microbial community compositions (Gao et al., PLOS ONE 21, e0344571 (2026)).

disordered (DIS)disordered (DIS)lamellar (LAM)lamellar (LAM)gyroid (GYR)gyroid (GYR)hexagonally packed cylinder (HEX)hexagonally packed cylinder (HEX)body-centered cubic (BCC)body-centered cubic (BCC)BCC–HEX boundary (Landau theory)HEX–GYR boundary (Landau theory)GYR–LAM boundary (Landau theory)DIS–BCC boundary (Landau theory)00.10.20.30.40.50.60.70.80.9100.511.522.53ϕχ
Phase diagram of diblock copolymer melts obtained from our neural network, trained on data below χ = 0.3 (yellow dashed line) and predicting the phase behavior above it. Here, ϕ and χ denote the volume fraction of block A and the Flory parameter, respectively. Dots: lamellar (LAM, red), gyroid (GYR, green), hexagonally packed cylinder (HEX, blue), body-centered cubic (BCC, magenta), and disordered (DIS, black). Curves: phase boundaries from the Landau theory of microphase separation, between GYR and LAM (red), HEX and GYR (green), BCC and HEX (blue), and DIS and BCC (magenta).
Ensemble OutputCombined Network OutputInput······NN1Output 1······NN2Output 2······NNnOutput n···
Ensemble neural-network architecture. The outputs of individual networks are combined into an ensemble output.
Machine-learning framework for predicting oil contamination in water
Predicting oil contamination in water from microbial community compositions: random-forest feature selection reduces 503 bacterial features to a few key ones, an augmented data neural network (ADNN) with noise injection expands the training data, samples generated by a variational autoencoder (VAE) serve as perturbations for stress testing, and a neural network trained on the augmented data predicts the oil type or absence (no oil, crude, or diesel). Feasibility demonstration: R2 up to 0.99 in training and stress testing, but a bottle-level hold-out mean R2 of −0.15, so validation on larger independent datasets is required.
Area 4 references
  1. Gao, T., Bigcraft, I., Techtmann, S., and Nakamura, I., “Predicting oil contamination in water using machine learning on microbial compositions”, PLOS ONE 21, e0344571 (2026).
  2. Gao, T., Shock, C. J., Stevens, M. J., Frischknecht, A. L., and Nakamura, I., “Surrogate molecular dynamics simulation model for dielectric constants with ensemble neural networks”, MRS Commun. 12, 966–974 (2022).
  3. Gao, T., Qian, Z., Chen, H., Shahbazian-Yassar, R., and Nakamura, I., “Inhibition of lithium dendrite growth with highly concentrated ions: cellular automaton simulation and surrogate model with ensemble neural networks”, Mol. Syst. Des. Eng. 7, 260–272 (2022).
  4. Nakamura, I., “Phase diagrams of polymer-containing liquid mixtures with a theory-embedded neural network”, New J. Phys. 22, 015001 (2020).