Underlayer material influence on electric-field controlled perpendicular magnetic anisotropy in CoFeB/MgO magnetic tunnel junctions

We study the dependence of the perpendicular magnetic anisotropy on the underlayer material in magnetic tunnel junction. Using several different 4d and 5d metals we identify an optimal seed layer in terms of high anisotropy, low mixing, and high thermal stability. In such systems we investigate the tunability of the anisotropy by means of electric fields. Especially, by using W as the underlayer of the CoFeB/MgO/CoFeB trilayer we could obtain good thermal stability that allows for annealing in the temperatures up to $450{}^{\circ }\mathrm{C}$, which results in high perpendicular anisotropy.

Figure 1

VSM measurements of the sample on the W buffer annealed at (a) $225{}^{\circ }\mathrm{C}$ and at (b) $350{}^{\circ }\mathrm{C}$ in both in-plane and perpendicular fields.

Figure 2

(a) Magnetic moment per unit area and (b) saturation field as a function of the annealing temperature for $M$/CoFeB/MgO sample with different underlayers.

Figure 3

Normalized resistance versus the in-plane magnetic field measured at different bias voltages on the sample with the W buffer annealed at $350{}^{\circ }\mathrm{C}$. The interface anisotropy energy is calculated for each bias voltage step.

Figure 4

Interface anisotropy energy change per applied electric field (a) and tunneling magnetoresistance (TMR) ratio (b) measured in MTJs with different buffers.

Figure 5

(a) Product of the magnetic moment and CoFeB layer thickness as a function of the nominal thickness $t$ and (b) product of the effective anisotropy and the effective thickness $t_{\mathrm{eff}}$ of the bottom CoFeB. Solid lines in (a) represent fits to the thicker CoFeB region, whereas dotted lines are fitted to the thinner CoFeB, for which the magnetic dead layer is determined.

Figure 6

Aging property measured for the MTJ with the W buffer annealed at $350{}^{\circ }\mathrm{C}$.

Figure 7

Interface PMA energy dependence on bias voltage after different annealing temperatures. Linear dependence of $K_{\mathrm{i}}$ versus $V_{\mathrm{b}}$ is maintained up to $450{}^{\circ }\mathrm{C}$ in case of the W buffer, whereas for Ir only up to $300{}^{\circ }\mathrm{C}$.

Figure 8

(a) FMR frequency peaks versus the magnetic field applied at ${60}^{\circ }$ from the sample plane for the sample with the W buffer annealed at $275{}^{\circ }\mathrm{C}$. Lower frequency signal is attributed to the bottom free layer (FL) and its frequency can be further tuned by the application of static bias voltage. Higher frequency signal originates from the top reference layer (RL). (b) FMR spectra obtained at the frequency of $f=12$ GHz under various bias voltages. Lines are artificially offset for clarity.