Effects of field annealing on MnN/CoFeB exchange bias systems

We report the effects of nitrogen diffusion on exchange bias in MnN/CoFeB heterostructures as a function of MnN thickness and field-annealing temperature. We find that competing effects occur in which high-temperature annealing enhances exchange bias in heterostructures with thick MnN through improved crystallinity, but in thinner samples this annealing ultimately eliminates the exchange bias due to nitrogen deficiency. Using polarized neutron reflectometry and magnetic x-ray spectroscopy, we directly observe increasing amounts of nitrogen migration from MnN into the underlying Ta seed layer with increased annealing temperature. In heterostructures with thin MnN layers, the resulting nitrogen deficiency becomes significant enough to alter the antiferromagnetic state before the Ta seed layer is nitrogen saturated. Furthermore, we observe intermixing at the MnN/CoFeB interface which is attributed to the nitrogen deficiency creating vacancies in the MnN layer after annealing in a field. This intermixing of Mn with Co and Fe is not believed to be the cause for loss of exchange bias when the MnN layer is too thin but is instead a secondary effect due to increased vacancies after nitrogen migration.

Figure 1

(a) Normalized MOKE measured with field applied along the in-plane field-annealing axis for 1.6 nm CoFeB on 48 nm (top) and 30 nm (bottom) MnN. (b) Exchange bias field (black, solid) and coercive field (green, dashed) for the 30 nm (solid) and 48 nm (open) MnN samples with 1.6 nm CoFeB.

Figure 2

Measured polarized neutron reflectivity (points) with theoretical fits (solid lines) in the non-spin-flip configuration. Results are shown for the 30 and 48 nm MnN in the as-deposited state, and after annealing at 325 and 525 °C. Error bars are representative of 1 σ.

Figure 3

Nuclear (top) and magnetic (bottom) scattering length density profiles for the 7 nm CoFeB samples with (a) 30- and (b) 48 nm of MnN. The layers are denoted by their nominal thicknesses and distance is referenced such that the substrate surface is at 0 nm.

Figure 4

XMCD and inset of XAS for (a) Fe, (b) Co, and (c) Mn with the XAS shown in the inset figures. (d) The $L_3$ peak intensity normalized to the preedge intensity as a function of annealing temperature. Data measured on the sample with 30 nm of MnN and 7 nm of CoFeB.