A surprising prediction of the new theory is that the experimentally observed polymorphism of the wetting liquid depends only on two parameters: (i) the channel geometry, i.e., the ratio of the channel depth to the channel width; and (ii) the interaction between substrate material and liquid. One has to distinguish seven different liquid morphologies which involve localized droplets (D), extended filaments (F), and thin wedges (W) at the channel corners. For microfluidics applications, the most important morphology regime is (F) which corresponds to stable filaments. Since this regime covers a relatively small region of the morphology diagram, it can only be obtained if one carefully matches the channel geometry with the substrate wettability. Thus, a water filament in a narrow channel that has a width of 100 nanometer can sustain an overpressure up to 15 atm. In contrast, if the channel had a width of one millimeter, the water filament could only sustain a thousandth part of an atmosphere.
One relatively simple application of the morphology is obtained if the system is designed in such a way that one can vary or switch the contact angle in a controlled fashion. One such method is provided by electrowetting; alternative methods, which have recently been developed, are substrate surfaces covered by molecular monolayers that can be switched by light, temperature, or electric potential.
The experiments described in the PNAS study use a polymeric liquid that freezes quickly and can then be scanned directly with the tip of an atomic force microsope. However, the same morphology diagram should also apply to other liquids and other substrate materials. It should also remain valid if one further shrinks the surface channels and, in this way, moves deeper into