“Vibration shakes the water out” is close enough to sound right and vague enough to mislead. The actual mechanics explain several things the summary does not.
For a complete overview, see what the frequency actually does.
Amplitude, not frequency
The common assumption is that faster vibration works better. It is the reverse. What moves
liquid is how far the diaphragm travels each cycle — its amplitude — and amplitude falls as
frequency rises. At 150 Hz a small driver moves a comparatively large distance. At
15,000 Hz the movement is measured in microns: audible, but doing no useful mechanical
work. This single relationship is why ejection tones are low and why tones advertised in the
tens of kilohertz cannot work.
The pumping action
Each outward stroke compresses air behind whatever sits in the cavity, nudging it toward the
opening. Each inward stroke does the reverse. Over a full cycle these roughly cancel, so what
produces net outward movement is gravity plus cavity geometry — which is precisely why grille
orientation changes results so noticeably.
Breaking surface tension
Before transport can happen, the film has to fragment. Water clinging to a mesh is held
together by cohesive force with a breaking threshold. Accumulated displacement over the first
few seconds exceeds it and the film breaks into droplets all at once. This threshold behaviour
is why water clears suddenly while dust clears gradually.
Why the phone's vibration motor does not help
A common question, and the answer is a clear no. The haptic motor vibrates the whole chassis
at low frequency with tiny displacement, and crucially it does not move the diaphragm relative
to the cavity. What matters is relative motion between the diaphragm and the trapped liquid.
Shaking the entire device moves both together and accomplishes nothing.
Why shaking the phone by hand is also poor
Same reason, plus a real risk: vigorous shaking can drive water further in through seams and
button wells rather than out through the grille. Gravity plus acoustic pumping is both safer and
more effective.
Why clipping reduces effectiveness
At maximum volume the waveform peaks flatten, so the diaphragm spends part of each cycle held
at full extension rather than travelling smoothly. The motion becomes a buzz rather than a pump.
Seventy to eighty percent preserves the clean sinusoidal travel that does the work.