A PN junction in equilibrium has a depletion region near the metallurgical junction and a built-in barrier potential that opposes the movement of majority carriers. When the junction is forward biased (p-side made positive with respect to the n-side), the applied voltage opposes and reduces this built-in barrier, narrowing the depletion region and allowing carriers to cross.

The primary current mechanism under forward bias is therefore:

• The reduced barrier lets majority carriers cross the junction: electrons from the n-side move into the p-side, and holes from the p-side move into the n-side.
• Once across, these carriers are now in a region where they are the minority — they become injected minority carriers.
• They then diffuse away from the junction through the neutral regions (down their concentration gradient), recombining with majority carriers as they go.

This injection and diffusion of minority carriers constitutes the forward (diffusion) current, so the correct description is majority carriers crossing the junction to become injected minority carriers that diffuse through the neutral regions.

The other options are physically wrong: forward current is dominated by diffusion, not by drift due to an enhanced barrier field — in fact forward bias reduces the barrier field, so the choice invoking an enhanced internal field is incorrect. Protons do not move in a semiconductor (they are fixed in the nuclei of the lattice), so no current arises from any "creation of protons." And carriers do not freeze or stop at the junction under forward bias — that would describe a blocked/reverse condition, the opposite of the forward-conduction behaviour being asked about.