Structured and viscous water in subnanometer gaps

Direct and simultaneous measurements of the normal and lateral forces encountered by a nanosize spherical silicon tip approaching a solid surface in purified water are reported. For tip-surface distances, $0\pm 0.03\mathrm{nm}<d<2\mathrm{nm}$, experiments and grand canonical molecular-dynamics simulations find oscillatory solvation forces for hydrophilic surfaces, mica and glass, and less pronounced oscillations for a hydrophobic surface, graphite. The simulations reveal layering of the confined water density and the development of hexagonal order in layers proximal to a quartz surface. For subnanometer hydrophilic confinement, the lateral force measurements show orders of magnitude increase of the viscosity with respect to bulk water, agreeing with a simulated sharp decrease in the diffusion constant. No viscosity increase is observed for hydrophobic surfaces.

Figure 1

(Color online) Left: an AFM was used to measure the normal and lateral forces between a nanosize untreated silicon tip and three different flat solid surfaces in de-ionized water. In this figure we also show a scanning electron microscopy (SEM) image of the tip apex. Right: an atomic configuration illustrating the MD simulation model. The water (oxygen in blue and hydrogen in red) and the quartz substrate (oxygen in yellow and silicon in dark yellow) are extended with the use of periodic boundary conditions in the $x\text{−}y$ plane (normal to the page). The tip is fully immersed in water.

Figure 2

$F_N$ vs $d$ for wetting [(A), $\left({\mathrm{A}}^{\prime }\right)$, (a), and $\left({\mathrm{a}}^{\prime }\right)$] and nonwetting [(b) and $\left({\mathrm{b}}^{\prime }\right)$] surfaces. The vertical dashed lines indicate the position of the force maxima corresponding to layer $n=1$, 2, and 3. In the inset of (a), we show values of $\delta$ corresponding to different layers, obtained from several measurements. The estimated error in $F_N$ is $\pm 0.05\mathrm{nN}$; the error in $d$ is $\pm 0.3\mathrm{\AA }$.

Figure 3

MD simulated water density profiles normal to the confining interfaces, shown for selected gap widths $(d=1.2$, 0.7, and $0.5\mathrm{nm})$ of the wetting (left) and nonwetting (right) confinements. The slight asymmetry in (c) between the two middle humps is statistical in nature.

Figure 4

Experimental $\eta$ vs $d$ as calculated from Eq. (1) (where $A=75{\mathrm{nm}}^2$ calculated for $\Delta h=0.25\mathrm{nm}$, see text) for (a) mica, (b) glass, and (c) HOPG. The estimated error in $F_L$ is $\pm 0.05\mathrm{nN}$; the error in $d$ is $\pm 0.3\mathrm{\AA }$. In the insets of these figures, we show for the corresponding surfaces the experimental $F_L∕v_{\mathit{\text{shear}}}$ vs $d$. (d) Simulated $D$ vs $d$ in water films confined by wetting and nonwetting interfaces. Only molecules remaining in the confinement during $150\mathrm{ps}$ are selected for computing the displacements. Note the nonoscillatory character of the viscosity and diffusion curves. Indeed, oscillatory variations of these quantities are found only when density layering in the gap is accompanied by the development of intralayer order (Ref. 27), which is not the case here.