Thermally Stable Perpendicular-Magnetic Tunneling Junction Spin-Valve Implementing W Bridge, Space and Cap-Layer

Abstract

Perpendicular spin-transfer-torque magnetic random access memory (p-STT MRAM) has been intensively researched because of the possibility of it overcoming the scaling limitations of current dynamic random access memory (DRAM) below the 10-nm design rule1. p-STT MRAM cells consist of a selective device and a perpendicular magnetic tunneling junction (p-MTJ) spin-valve. Moreover, a lot of research has gone into improving three device parameters of these devices, i.e., the tunneling magnetoresistance (TMR) ratio, thermal stability (Δ), and switching current (JC ), at the back end of line (BEOL) temperature of 400oC3. Note that the BEOL temperature of 400oC is required in order to integrate memory cells3. Tantalum (Ta) has been used as a bridge and spacer material to ferro-couple two ferro-magnetic layers. However, so far, it has proven extremely difficult achieve a high TMR ratio in a MgO based p-MTJ spin-valve using a thickness Ta spacer at the BEOL temperature of 400oC4. In our experiment, we tried a new spacer material with a body-centered-cubic (bcc) crystal structure (i.e., tungsten). We investigated the dependency of the TMR ratio on the thickness and type of material (tantalum or tungsten) of spacers in double MgO-based p-MTJ spin-valves with a top free Co2Fe6B2 layer. Two types of double MgO based p-MTJ spin-valves with a [Co/Pt]n-based SyAF layer were fabricated on a 12-inch-diameter wafer deposited with SiO2/W/TiN films in a 12-inch multi-chamber sputtering system under a high vacuum of less than 1×10−8 torr (without breaking the vacuum; see Figs. 1a and b). The thicknesses of the Ta and W spacers were varied between 0.2 and 0.7 nm. All samples were subject to ex-situ annealing at 400oC for 30 min under a perpendicular magnetic field of 3 Tesla. The dependence of the TMR ratio on the thickness and type of material of the spacer was estimated by using current-in-plane tunneling (CIPT) at room temperature, as shown in Fig. 1c. For the Ta spacers, the TMR ratio significantly increased from ~38 to ~95% as the spacer thickness was increased from 0.20 to 0.35 nm. It remained about 98% for spacer thicknesses ranging from 0.40 to 0.53 nm. It decreased from ~90 to ~42% as the spacer thickness was further increased from 0.58 to 0.70 nm. Thus, the TMR ratio (~98%) peaked in a certain spacer thickness region (0.40~0.53 nm) for Ta. Next, regarding the samples made with W spacers, the TMR ratio significantly increased from ~85 to ~131% as the spacer thickness was increased from 0.20 to 0.30 nm. It remained at ~134% as the spacer thickness was further increased from 0.40 to 0.53 nm. The ratio slightly decreased from ~134 to ~111% for larger thicknesses ranging from 0.58 to 0.70 nm. Thus, the TMR ratio (~134.0%) also peaked in a specific spacer thickness region (0.40~0.53 nm) for W. Comparing the TMR ratios in the cases of the Ta spacer and the W spacer obviously indicates that the use of 0.20 and 0.70 nm-thick W spacers resulted in least 35% higher TMR ratios in the comparison with the use of Ta spacers. In particular, both the thinnest (0.20 nm) and thickest (0.70 nm) W spacers yielded much higher TMR ratios (~85 and ~111%) in comparison with the corresponding Ta spacers (~38 and ~42%). In our presentation, we will report the mechanism by which thickness and material of the spacer influence the TMR ratio was revealed by examining the static magnetization behavior, (100) bcc crystallinity, and depth profile of the atomic composition of the spin-valves. In addition, we will present the result of effects on W bridge/cap-layer compare with Ta bridge/cap-layer in Double MgO based p-MTJ spin-valve. * This work was supported by a Basic Science Research Program grant from the National Research Foundation of Korea (NRF) funded by the Korean government (MSIP) (No. 2014R1A2A1A01006474) and the Brain Korea 21 PLUS Program in 2014. Reference [1] K. C. Chun et al., IEEE J. Solid-st. Circ. 48, 598 (2013) [2] K. Yamane et al., IEEE T. Magn. 49, 4335 (2013). [3] L. Thomas et al., J. Appl. Phys. 115, 172615 (2014) [4] J. Swerts et al., Appl. Phys. Lett. 106, 262407 (2015) Figure 1