Study on spin-orbit torque based non-volatile logic-in-memory applications

Abstract

Spintronics is a study of electron spin, which can is a promising candidate for future transistors. A recentlydiscovered spin-orbit torque (SOT), is a new topic in spintronics, which allow the switching of magnetization byin-plane current through heavy metal (HM)/ferromagnet (FM)/oxide structure. SOT based switching is known forits high speed and high reliability. To enhance the efficiency of the SOT device, HM material with large spinHall angle is needed. On the other hand, over all these advantages, SOT devices require external in-planemagnetic field for deterministic switching, which is detrimental for device applications.This thesis focuses on the development of non-volatile spintronics logic device based on SOT. SOT-baseddevices are good candidates for logic-in-memory applications. For the given purpose, first a material search ofheavy metals with large spin Hall angle is performed. Prior reports suggest that tungsten (W) may have a largespin Hall angle compared to that of Ta or Pt, which are well known heavy metals that can induce perpendicularmagnetization with either CoFeB or Co. Moreover, as Ta and W are CMOS compatible material, W is studied asa replacement for Ta for its large spin Hall angle. Here, I observed that W is more SOT efficient than that of Ta.Then, a search for new underlayer material is performed. New underlayers such as anti-ferromagnets canbring new phenomena such as field-free switching. In this point of view, I introduce ferromagnetic material asthe underlayer. Even though 3d metals are known to have negligible spin-orbit coupling and SOT, some reportsshow that ferromagnetic metals have effective spin Hall angle, enough to produce non-negligible SOT. Thus,introducing a thin layer of FM as the source of SOT can allow a new degree of freedom on the development ofSOT-based devices.Finally, a prototype of logic-in-memory device based on spin-orbit torque is developed. A 2-bit logic-inmemorydevice is demonstrated which shows that by controlling the anisotropy of each bit, a selective switchingto simultaneous switching can be achieved. To control each bit independently, I introduced a gate oxide to thestructure, thereby enabling external application of electric field on the ferromagnetic layer. It has been reportedthe magnetic properties can be controlled by applying voltage on it. Thus, in this scheme, I applied externalvoltage as gate voltage and controlled the switching current of each FM bit. This allowed a full control of 2-bitlogic which was unable in the prototype of the device.