This mechanism involved mainly an adatom and 2 atoms in the upper substrate layer. After passing through the saddle configuration, the adatom was found 2 rows down and 1 row over from its original position. The movement of the adatom was therefore like that of a knight on a chessboard. The activation barrier for this mechanism was high, and it began to appear at homologous temperatures of about 0.5. Due to the long-range nature of this mechanism, it could significantly affect the diffusion rate as compared to the typical hopping between adjacent minima or diffusion via the exchange mechanism.

J.M.Cohen: Surface Science Letters, 1994, 306[1-2], 545-9