Electron Paramagnetic Resonance Measurements of Four Nitroxide Probes in Supercooled Water Explained by Molecular...

Electron paramagnetic resonance (EPR) measurements of the rotational diffusion of small nitroxide probes have been demonstrated to be a powerful technique for experimentally investigating the properties of supercooled liquids, such as water. However, since only the rotational diffusion of the probe molecules is measured and EPR measurements are indirect, it is not clear what the relationship between the behavior of water and the probe molecule is. To address this, we have performed molecular dynamics simulations of four nitroxide probes in TIP4P-Ew and OPC water models to directly compare with EPR experiments and to determine the behavior of the water and the underlying microscopic coupling between the water and the probes. In all, 200 ns simulations were run for 23 temperatures between 253 and 283 K for all four probes with each water model for an aggregate of 36.8 μs of simulation time. Simulations for both water models systematically underestimated the rotational diffusion coefficients for both water and probes, though OPC simulations were generally in better agreement with the experiments than TIP4P-Ew simulations. Despite this, when the temperature dependence of the data was fit to a power law, fit parameters for TIP4P-Ew were generally in better agreement with the experiments than OPC. For probe molecules, the singular temperature was found to be T₀ = 226.5 ± 0.4 K from experiments, T₀ = 208 ± 2 K for OPC water, and T₀ = 215 ± 2 K for TIP4P-Ew water. While for water molecules, the singular temperature was found to be T₀ = 220.3 ± 0.2 K from experiments, T₀ = 208 ± 2 K for OPC water, and T₀ = 220 ± 1 K for TIP4P-Ew water. Systematic underestimation of the rotational diffusion coefficients was most pronounced at lower temperatures and was clearly observed in changes to the Arrhenius activation energy. Above the maximum density temperature of T_ρ = 277 K, an activation energy of E_A ≈ 16.7 kJ/mol was observed for the probes from experiments, while OPC had E_A ≈ 15.2 kJ/mol and TIP4P-Ew had E_A ≈ 14.6 kJ/mol. Below the maximum density temperature, the activation energy jumped to E_A ≈ 32.5 kJ/mol for experiments but only E_A ≈ 23 kJ/mol for OPC and E_A ≈ 22 kJ/mol for TIP4P-Ew. In all cases, we saw good agreement between the behavior of the probe molecules and water. To understand why, we calculated the average number of hydrogen bonds between the probe molecules and water. From this, we were able to explain the rotational diffusion times for all of the probes. These results show that current molecular models are sufficient to capture physical phenomena observed with EPR and to help elucidate why the probes provide accurate insights into the behavior of supercooled water.