Perhaps what you are referring to is the no-slip condition, in which fluid at a solid boundary will have a macroscopic velocity relative to the boundary of $\mathbf{0}$.
I've seen two arguments for it:
- Adhesion beats cohesion near the boundary, and the particles near the boundary will be adhesively joined to the boundary.
- Particle collisions against the non-specular boundary and neighboring particles will probabilistically trap particles near the boundary.
This second argument provides a nice explanation for why this condition fails in rarefied gas flows, namely that particle-particle collisions are not frequent enough to continually push particles at the boundary back if they try to escape.