Effect of ultra-thin Cu underlayer on the magnetic properties of NisoFe2, / FeVMn, 0 films C. LIU, L. SHEN, H. JIANG, D. YANG, G. WU, C. ALEXANDER, AND G. J. MANKEY Center for Materials for Information Technology, University of Alabama, Box 870209, Tuscaloosa, AL 35487-0209, gmankey@mint.ua.edu ABSTRACT The NijFe 20 / Fe_,Mn_,thin film system exhibits exchange bias behavior. Here a systematic study of the effect of atomic-scale thin film roughness on coercivity and exchange bias is presented. Cu (t) / Ta (100 A) / Ni.Fe 20 (100 A) / Fe,0Mno (200 A) / Ta (200 A) with variable thickness, t, of the Cu underlayer were DC sputtered on Si (100) substrates. The Cu underlayer defines the initial roughness that is transferred to the film material since the film grows conformal to the initial morphology. Atomic Force Microscopy and X-ray diffraction were used to study the morphology and texture of the films. Morphological characterization is then correlated with magnetometer measurements. Atomic Force Microscopy shows that the root mean square value of the film roughness exhibits a maximum of 2.5 A at t = 2.4 A. X-ray diffraction spectra show the films are polycrystalline with fcc (111) texture and the FeoMn, 0 (111) peak intensity decreases monotonically with increasing Cu thickness, t. Without a Cu underlayer, the values of the coercivity and loop shift are, H, = 12 Oe and Hp = 56 Oe, respectively. Both the coercivity and loop shift change with Cu underlayer thickness. The coercivity reaches a maximum value of H, = 36 Oe at t = 4 A. The loop shift exhibits an initial increase with t, reaches a maximum value of HP = 107 Oe at t = 2.4 A, followed by a decrease with greater Cu thickness. These results show that a tiny increase in the film roughness has a huge effect on the exchange bias magnitude. INTRODUCTION Meiklejohn and Bean discovered Exchange anisotropy in 1956 [1]. Its origin is the interfacial exchange coupling between a ferromagnetic material and an antiferromagnetic material. Although extensive work has been done recently due to the application of exchange biasing in MR read heads and MRAM, the mechanism of the effect is still not well understood. For example, the magnitude of biasing field that is predicted by theory is too large compared to experimental results. Malozemoff proposed a model [2] for a compensated interface, considering the randomness in exchange interactions at the interface arising from surface roughness or chemical inhomogeneity on an atomic scale. This model can predict the correct order of biasing field. Detailed experiments must be performed to test the theory. Generally, interface disorder has profound and complicated effects on exchange bias. Disorder can be generated by roughness, variation in crystallinity ( i.e. the degree of texture ) or grain size. Contributions from these factors are usually difficult to distinguish. Studies of the effect of interfacial roughness on the exchange coupling and how it influences the magnetic properties of the ferro- antiferromagnetic system have identified some problems [3, 4, 5]. One may change the roughness at the interface and investigate its effect on magnetic properties of the pinned layer. A recent work by Hwang et al. [3] showed biasing field and coercivity of pinned NisoFe 20 of NiO/Ni 80 Fe 20 bilayer can be greatly influenced by the interfacial roughness introduced by a very thin Si 3 N 4 buffer layer. They found an increase in biasing field due to roughness. However, in the case of FeMn as the antiferromagnetic material, a contradictory conclusion was reached [4, 69 Mat. Res. Soc. Symp. Proc. Vol. 562 01999 Materials Research Society