Hydrogen Bonding Inside Anionic Polymeric Brush Layer: Machine Learning-Driven Exploration of the Relative Roles of the Polymer Steric Effect, Charging, and Type of Screening Counterions

This paper employs a combination of all-atom molecular dynamics (MD) simulations and unsupervised machine learning (ML) for studying the water-water hydrogen bonds (HBs) inside the anionic poly acrylic acid (PAA) brushes modeled using all-atom MD simulations. PAA brush layer with different charge fraction (f), namely, f = 0, 0.25, and 1, is considered. Water-water interactions, both inside and outside the brush layer, are represented through distinct clusters of tupules of variables representing distances associated with the interacting water molecules. While clusters representing the HBs are present for water inside and outside the brushes, several clusters representing the long-range water-water interactions are missing for the water molecules inside the highly charged (f = 1) PAA brushes. More importantly, inside highly charged brushes, the edge of the clusters representing the water-water HBs is progressively shortened as compared to that in the bulk. Both of these results stem from the presence of the PAA brushes imparting the steric effect and the charge effect, or the effect associated with enhanced interactions of water molecules with PE charges and counterions, thereby disrupting the water connectivity. This water-charged-species interaction also increases the water-water HB angle, i.e., makes the water-water HBs less stable inside the highly charged PAA brush layer. The narrowing of the clusters representing the HBs and the alteration of the angle characterizing the HBs confirm that the conditions defining the water-water HBs change inside the PAA brush layer as a function of the charges on the PAA brush layer. Furthermore, we show that the use of the generic definition of HBs, as compared to using our simulation-motivated modified definition of water-water HBs, overpredicts the number of water-water HBs inside the PAA brush layer. Finally, we employ this all-atom-MD-ML framework to quantify the effect of other types of screening counterions (Li+, Ca2+, and Y3+ ions) in determining the water-water interactions and water-water HB properties inside the PAA brush layer. The findings of the present study, confirming the weakening of water-water HBs inside the PAA brush layer, point to the possibility that the water molecules will be more available for hydrating the brush layer and counterions, thereby leading to a more pronounced wetting of the PAA brush layer.