Macro model for stochastic behavior of resistance distribution of magnetic tunnel junction

Abstract

In this work, we fabricated MgO-based magnetic tunnel junction (MTJ) samples to observe behavior of resistance variation, and investigated a stochastic behavior model for MTJ resistance from measured real data. We found the relationship between parallel resistance (R P ), anti-parallel resistance (R AP ), and TMR from the measurements. The variation of barrier thickness affects not only resistance but also TMR. This means that broad R AP distribution is caused by R P distribution. In addition, R AP distribution can be reduced by increasing temperature and bias voltage. We developed a macro model that can evaluate resistance distribution based on the stochastic behavior of MTJ resistance variation from only t ox varied. The amount of resistance variation, which is considered with regard to the circuit performance, can be obtained from Δt ox designed by designer. In addition, the impact for operating circumstance such as bias and temperature can be considered by using fit equations.

Stochastic macromodel of magnetic tunnel junction resistance variation and critical current dependence on resistance variation for SPICE simulation

“…28,29) The asymmetric characteristics have not been clearly defined theoretically, and we assumed the ratio between η AP and η P to be as described by Eq. ( 6), where 0.6 is obtained for our 17) simulation by fitting the results obtained using the P-MTJ sample introduced by Suzuki et al 30)…”

“…[19][20][21] On the basis of the obtained results, we proposed an MTJ macromodel that realizes the resistance variation in accordance with t ox , A, T, and V b variations. 17) To realize this macromodel, ideal parallel resistance should be obtained using Eq. ( 1), which is derived from Brinkman's tunneling conductance.…”