Shape-engineered ferromagnets and micromagnetic simulation techniques for spin-transfer-torque random access memory
Spin-transfer-torque random access memory (STTRAM) has received great attention as a prospective universal memory due to high speed read and write capabilities, scalability to smaller technology nodes and non-volatile data retention. Two major factors that could limit the performance of large scale STTRAM arrays are the high switching current and the stochastic switching behavior. In this work, possible routes to mitigate these issues have been explored and new techniques have been proposed to estimate the reliability of the write process. Large area of the selection transistor required to support high switching current impacts the bit storage density of an STTRAM memory array. To increase the bit storage density, a multi-state STTRAM cell employing a cross-shaped ferromagnet was proposed previously. Here, the spin-transfer-torque (STT) driven mag-netization dynamics of the cross-shaped ferromagnet is revisited. As a low power alternative, voltage controlled magnetic anisotropy (VCMA) based writing scheme is studied. Trade-offs and limitations of the VCMA-induced switching over STT are also discussed. In the next part of this dissertation, magnetic properties and magnetization process of epitaxial chromium telluride thin films have been studied. Presence of strong perpendicular magnetic anisotropy in this material makes it an attractive choice for device applications. In this work, anisotropy energies of chromium telluride thin films have been estimated from magnetization measurements. The magnetization reversal process is then studied using analytical models as well as micromagnetic simulations. The last part of this work focuses on the write error rates (WER) of STTRAM. The stochastic write process of STTRAM at finite temperatures gives rise to write errors when a bit fails to switch within the duration of the write pulse. Ultra-low WER on the scale of 10⁻⁹ or less are desired for practical applications. Micromagnetic simulations are required to capture spatially-incoherent magnetization dynamics inside a ferromagnet, which may effect the WER. In this work, using the techniques of rare event enhancement, reliable calculation of WERs to 10⁻⁹ is demonstrated while keeping the computational effort to a minimum. Employing rare-event-enhanced micromagnetic simulations, WERs of both perpendicular and in-plane STTRAM bits are calculated and effects of spatially-incoherent excitations on the WER slopes are discussed.