Recent progress in reducing the current and time for magnetization switching in magnetic tunnel devices for memory applications

Invited

Abstract

In spin-transfer-torque switched magnetic random access memory (STT-MRAM) using magnetic tunnel junctions (MTJ), the STT switching current directly affects selection transistor size and circuit density. This is a critical attribute for cost competitiveness in commercial applications. Recent developments in STT-MRAM have been focusing on reducing switching current well below 100 uA, increasing speed beyond 50 ns, and improving switching reliability to better than 10-9, while simultaneously aiming for data retention of 10 years at operating temperatures ranging between -40 and + 85C or beyond. In this talk, I will review the fundamental device and materials physics that govern the STT-switching dynamics. While the switching threshold current is more directly related to data-retention, the actual operating switching current for ensuring a certain speed and reliability (switching error) for fast, below 10ns switching can also exceed macrospin expectations. Experiments show the switching current required to achieve deep error floor can be less than estimated from macrospin. For further improvements it is important to understand the dynamics involving a nanomagnet's internal degrees of magnetic freedoms during STT switching, a potentially complex subject rich in nonlinear dynamics. These complex nonlinear mechanisms can speed-up the intended switching of the MTJ, but can also affect the reference-layer stability, requiring special care. Longer term, for even faster manipulation of nanomagnets approaching or below 1 ns, new sources of spin-current with better charge-to-spin current conversion needs to be considered, such as thermal magnonic or spin-orbit-derived spin-currents. I will briefly discuss recent advances with these new sources of spin currents, and the likely common challenges they will give rise to, in terms of materials and device design and development.

Presenters

  • Jonathan Sun

    IBM T. J. Watson Research Center, IBM Thomas J. Watson Research Center

Authors

  • Jonathan Sun

    IBM T. J. Watson Research Center, IBM Thomas J. Watson Research Center