AmericanChemicalSociety.com

The transition from microscale to nanoscale fluidic devices increases the effect of the device surface as the surface to volume ratio increases, which is both a distinct advantage and disadvantage. The advantages include the ability to manipulate individual biomolecules and create “lab on a chip” or “micro total analysis” systems, while the disadvantage is that such strong interactions between device surfaces and solutes can foul device performance and render the device unusable. In order to employ nanoscale fluidic devices in practical applications, it will be necessary to have the capability to predict the flow of solutes through nanochannels to assess the risk of clogging or fouling. Similar challenges arise when oil is hydraulically extracted from an oil-sand mixture, where the contact between the sand grains can be considered as a nanogap.

Unfortunately, there is no standard method to determine how individual molecules will interact with a given nanostructure. Predicting and interpreting device behavior from first principles is difficult, as it requires precise knowledge of every interaction between the device surface and atoms of the solute and solvent. Molecular dynamics (MD) offers a unique solution to this problem, as it accurately describes the motion of atoms on the length and time scales required for complete characterization of surface-solute interactions. However, MD alone cannot describe processes on the length and time scales of the actual device.

Here we report the results of an ongoing theoretical and computational study that aims to develop an accurate continuum model of the flow and sorption/desorption of small solutes in nanochannels and nanogaps derived from the atomic structure of the solute and the device surface. Our model is based on all-atom MD simulations examining sorption/desorption of a prototypical solute onto several nanochannel surfaces as a function of surface roughness, hydrophobicity and flow velocity. The model is parameterized by calculating the three-dimensional potential of mean force for a solute in a nanochannel and the solute diffusivity. Preliminary results reveal that our model predicts the correct behavior for solutes in featureless nanochannels, but there may be solute concentration-dependent effects for atomically rough surfaces, which require further investigation. Once complete, we expect our model to be instrumental in designing complex and automated nanofluidic systems and methods for oil extraction from oil-sand deposits.

To design an accurate continuum model of solute transport through nanofluidic devices, we first investigated the flow of water and solutes through a model nanochannel structure using all-atom MD. To determine how atomic-scale interactions affect water flow and solute sorption/desorption, we created three atomic-scale models of silica nanochannels measuring 5.5 nm in height. These systems were constructed to have varying surface roughness and hydrophobicity. A fourth nanochannel was created to have a smooth, frictionless surface, enabling us to study solute sorption/desorption without the effects of atomic-scale roughness. We have chosen dimethyl-methylphosphonate (DMMP), a small polar molecule, as a representative chemical solute. We have performed over 1 microsecond of MD simulations of these nanochannels and have characterized the behavior of water and solute in each. Such analysis served as a basis for deriving and calibrating our continuum transport model, This model decomposes solute transport into diffusion with diffusivity D, drift along the water flow with velocity v(x), and diffusion in a potential F(x), which determines the adsorption to the surface.

To parameterize our continuum model, we used the umbrella sampling method to compute the three-dimensional potential of mean force (PMF), F(x), of a DMMP molecule in each nanochannel. Independently, we computed a concentration-dependent diffusion constant, D, of DMMP in water through a set of MD simulations. For the smooth, frictionless nanochannel, our continuum model accurately predicts the transient and equilibrium behavior of DMMP in flow. Silica nanochannels present a more difficult case, as the PMF varies with position over the surface, and our simple model that was derived in the zero-concentration limit fails to predict the correct behavior for DMMP in the flow. Future work will extend our continuum model to include orientational and cooperative binding effects to account for phenomena seen in all-atom simulations of solute flow through a nanochannel.