Controlling water-mediated interactions by designing self-assembled monolayer coatings

Engineered nanoparticles have been broadly used in biological and geological systems. Hydrophilic molecules such as polyols have been used as coatings on nanoparticle surfaces due to their good biocompatibility and solubility in saline water. However, polyol coatings can cause huge retention of nanoparticles when encountering mineral surfaces. Here, molecular dynamics simulations enlightened that the strong adhesion of hydrophilic coatings to mineral surfaces stemming from the partitioning of the hydroxy groups on the hydrophilic molecules to the well-defined bound hydration layers on the mineral surfaces. To mitigate the nanoparticle adhesion, we investigated introducing small percentages of omniphobic fluoroalkanes to form a bicomponent system of hydrophilic and fluoroalkanes, which greatly perturbed the hydration layers on mineral surfaces and resulted in nonstick surface coatings. Our results provide important insight for the design of tunable “stickiness” nanoparticle coatings in different mineralogies, such as applications in subsurface environments or targeted delivery in mineralized tissues.

Designing self-assembled monolayers with tunable adhesions on calcite surfaces. Figure 2d shows the potential of mean forces (PMF) for the C8OH-C2F6 SAM interactions on calcite surfaces as integrated from Fig. 2a with different percentages of fluoroalkanes. As shown, the bicomponent SAM become purely repulsive at > 40% mole percentages of fluoroalkanes. As established in the previous section that the fluoroalkanes can perturb the hydration layers on the calcite surfaces and reduce the SAM adhesion, we envisioned that the more fluorinated alkanes should have more pronounced effects. This is clearly observed as in Fig. 2b,c,e,f, which show the pressure-distance curves and PMF for the C8OH-C2F8 and C8OH-C2F10 SAM interactions on calcite surfaces. As shown in Fig. 2e,f, the bicomponent C8OH-C2F8 and C8OH-C2F10 SAM become purely repulsive at > 20% and > 10% fluoroalkanes, respectively. Finally, we investigated the interactions between calcite surfaces and the SAM consisting of either only a more hydrophilic C8OH3 or bicomponent C8OH3-C2F10 (Fig. 5). For the pure alkanol (C8OH3) SAM (0% www.nature.com/scientificreports/ fluoroalkane C2F10), the pressure-distance curve in Fig. 5a shows a weak adhesion distance at d = 2.5 nm that corresponds to the partitioning of C8OH3 hydroxy groups to the third hydration layer on calcite (Fig. 6a), and a strong adhesion distance at d = 2.26 nm that corresponds to the partitioning of C8OH3 hydroxy groups to the first and second hydration layers on calcite (Fig. 6b). Note C8OH3 has multiple hydroxy groups on a single molecule (c.f. Figure 1c) that can concurrently participate in multiple hydration layers on calcites surfaces (Fig. 6b). In contrast, C8OH has only one hydroxy group on a single molecule that can only participate selectively in single hydration layer (

Discussion
Water-mediated interactions have been intensively studied for many decades 27 . Nevertheless, even though they are ubiquitous in all biological and technological systems, a unifying physical picture has yet to be achieved, and many hydration interactions were considered very surface specific [28][29][30] . In this work, we study the hydrophilic mineral surfaces, which are a class of very important yet less explored surfaces for hydration interaction. The hydrophilic SAM-mineral interactions could be viewed as nonsymmetric interactions between two hydrophilic surfaces 31 ; however, caution should be taken that they may not be understood by merely considering the competition of surface-surface and surface-water interactions, but rather how the highly well-defined water structures on the mineral surfaces (as far as the third hydration layers, see Figs. 5 and 6) may be complied or perturbed, which, we believe, is a unique feature of the hydration forces on mineral surfaces. Finally, we want to stress that the water-mediated interaction is a cooperative process (i.e., nonadditive) 32,33 , as demonstrated here that only 10% fluoroalkane could completely invert the SAM from adhesive to repulsive (c.f. Figs. 2f and 5b). This character has great implications in scaling-up SAM coatings for industrial applications since fluorinated molecules are usually much more expensive compared to the simple hydrophilic alkanols. Reducing the usages of expensive fluorinated molecules while maintaining the desired properties of the SAMs provide immense economic advantages 34,35 .
In summary, in this work using atomistic MD simulations we have studied in detail the water-mediated bicomponent hydrophilic/fluoroalkane SAM interactions on calcite surfaces. We first established that the molecular origin of the strong adhesions of the hydrophilic alkanols on the calcite surfaces were due to the partitioning of the hydroxy groups on the alkanols to the well-defined hydration layers on the calcite surfaces. We then www.nature.com/scientificreports/ investigated the effects from different types and compositions of fluoroalkanes and found that only 10% of the more fluorinated alkanes (C8OH-C2F10 and C8OH3-C2F10 SAMs) were enough to convert the SAMs to be purely nonstick on calcites. Even though the molecular details put forward in this work have been very helpful in designing the bicomponent surfaces as promising nanoparticle coatings with variable mineral adhesions, further experiments are needed to validate our predictions.

Methods
The model system consisted of SAMs made with mixtures of hydrophilic alkanols and fluoroalkanes on calcite (1 0 1 4) surfaces separated with distances d (Fig. 1a). Linear alkanols CH 3 (CH 2 ) 7 OH and CH 3 (CH 2 ) 4 (CHOH) 2 CH 2 OH (denoted C8OH and C8OH3; Fig. 1c) were used as hydrophilic molecules, and fluoroalkanes CH 3 CH 2 (CF 2 ) 5,7,9 CF 3 (denoted C2F6, C2F8, and C2F10; Fig. 1c) were used to modulate the SAM adhesion on calcite surfaces. The last carbons (on the opposite sides of the hydroxy or fluoro groups) of the bicomponent system were grafted onto a hexagonal lattice with density = 3.8 molecules/nm 2 that resembled experiment values 36 , and soft restraints were used on the x and y directions if the molecules exceeded the borders of the outermost grating lattices (Fig. 1b). This design (the position restraint and flat-bottomed position restraint algorithms as implemented in GROMACS software) 37 ensured the structural integrity of SAM while enabled the proper samplings of the water-mediated SAM-calcite interactions 38 . The calcite, calcite-water, and calcite-SAM molecule interactions were described by the force field of Xiao et al. 39 in which the hydrophilic and fluoroalkanes were described by the OPLS-AA 40,41 force field, and the water was treated with the TIP3P 42 model. The geometric combination rule was used to deduce all pairwise 12-6 Lennard-Jones potentials from atom-wise CaCO 3 , OPLS-AA, and water model parameters except for the H-F interactions, which were further optimized to reproduce experimental alkane and perfluoroalkane mixing properties 43 . This force field combination has been used to study other organic molecule-calcite interactions in solution with good compatibility [44][45][46] . www.nature.com/scientificreports/ Series of MD simulations with different SAM-calcite distances d were carried out to obtain the pressuredistance curves under NPT condition with T = 300 K and P = 1 bar. For each d, the SAM with area ~ 3 × ~ 3 nm 2 was placed at the designed height above an extended calcite slab perpendicular to the z axis in a simulation box with initial box size 5 × 5 × 5 nm 3 and periodic boundary conditions in three dimensions (Fig. 1a). The simulation boxes with predefined SAM and calcite surfaces were then solvated by 2400 water molecules. For each pressure-distance curve, we performed simulations with d = 2.70 to 2.42 nm with 0.04 nm increments followed by d = 2.40 to 2.10 nm with 0.02 nm increments. The simulation times for the MD runs were 40 ns for d > 2.40 nm and 100 ns for d ≤ 2.40 nm, with data analysis performed for the last 20 ns. Statistical uncertainties for the force measurements at each d were acquired by performing 3 independent simulations with randomized initial configurations of the SAMs.
At each SAM-calcite distance d, the pressures on SAM were measured from MD trajectories. At each MD time step, we first summed the total forces exerted on all the atoms that construct the SAM, and then calculated by dividing the sum of total forces on SAM by the projected area of SAM on the x-y plane to give the pressures.
Finally, the potential of mean forces (PMF) were calculated by PMF(d) = ∞ d � d ′ dd ′ , where Π(d) were the pressures measured at each d as described above (i.e., pressure-distance curves) and we used the max distance d = 2.7 nm as the reference state 19,33 .
All simulations were carried out using GROMACS v2018.2 37 . Electrostatic interactions were calculated using the particle-mesh Ewald summation, with a real-space cutoff of 1 nm, a grid spacing of 0.16 nm, and fourth-order interpolation. The van der Waals and neighbor-list cutoffs were both set to 1 nm. We used velocity rescaling temperature coupling with a time constant of 0.5 ps and Berendsen semi-isotropic pressure coupling with a  www.nature.com/scientificreports/ time constant of 5 ps. The simulation time step was set to 2 fs. The bottom 2 layers of the calcite molecules were restrained (using GROMACS position restraint algorithm) 37 in order to support the surface, while the top 2 layers of the calcite molecules facing the SAM could freely move 45,46 . All bonds were constrained using the LINCS algorithm 47 except for water molecules, which were constrained using the SETTLE algorithm 48 .

Data availability
The data sets generated and analyzed during the current study are available from the corresponding author on reasonable request.