AVS 46th International Symposium
    Magnetic Interfaces and Nanostructures Technical Group Wednesday Sessions
       Session MI-WeA

Paper MI-WeA3
Surface Diffusion Mechanism for the Exchange Coupling between a Ferromagnetic Layer and an Antiferromagnetic Layer

Wednesday, October 27, 1999, 2:40 pm, Room 618/619

Session: Giant Magnetoresistance
Presenter: C. Hou, The University of Alabama
Authors: C. Hou, The University of Alabama
K. Zhang, The University of Alabama
T. Zhao, The University of Alabama
H. Fujiwara, The University of Alabama
Correspondent: Click to Email
It is the surface net spin that is considered to be responsible for the exchange coupling between a ferromagnetic (F) layer and an antiferromagnetic (AF) layer. For an AF surface with roughness, statistics shows that an AF grain with a total number of spins n = n@super +@ + n@super -@, has an average net number of spins pointing in one direction of n@sub net@ = absolute value of (n@super +@ - n@super -@) = n@super 1/2@ with n@super +@ and n@super -@ denoting the number of spins in "+" and "-" directions within one AF grain, respectively. For an AF layer with total grains N = N@super +@ + N@super -@, it is expected N@super +@ /N@super -@ = 1 as deposited without the F layer, where N@super +@ and N@super -@ denote the number of grains with net spin in the "+" and "-" directions, respectively. Therefore the surface spins of all the AF grains still cancel each other. When a F layer is deposited on top of the AF layer under a field in "+" direction, the balance of N@super +@ and N@super -@ is broken, resulting in N@super +@ / N@super -@ > 1. Thus, the overall net moments are obtained without changing n@sub net@ of each grain. It is generally accepted that this is the cause of the exchange coupling. A NiFe/FeMn/NiFe sample shows reasonable amount of exchange bias field H@sub eb@ and coercivity H@sub c@ for both of the NiFe layers as deposited in a field. With post annealing with a field in the pinned direction, it is found that both H@sub eb@ and H@sub c@ of the two NiFe layers are increased. The results can't be explained by the increase of only the ratio N@super +@ /N@super -@. It is concluded that it is the increase of the individual n@sub net@ that causs the above phenomenon. The increase of n@sub net@ is thought to occur through surface spin diffusions.