Optimization of Tungsten $\ensuremath{\beta}$-Phase Window for Spin-Orbit-Torque Magnetic Random-Access Memory

Optimization of Tungsten $β$ -Phase Window for Spin-Orbit-Torque Magnetic Random-Access Memory

Kiran Kumar Vudya Sethu, Sambit Ghosh, Sebastien Couet, Johan Swerts, Bart Sorée, Jo De Boeck, Gouri Sankar Kar, and Kevin Garello

Phys. Rev. Applied 16, 064009 – Published 3 December 2021

Abstract

Switching induced by spin-orbit torque (SOT) is being vigorously explored, as it allows the control of magnetization using an in-plane current, which enables a three-terminal magnetic-tunnel-junction geometry with isolated read and write paths. This significantly improves the device endurance and the read stability, and allows reliable subnanosecond switching. Tungsten in the $β$ phase, $β$ - $W$ , has the largest reported antidamping SOT charge-to-spin conversion ratio $(θ_{AD} \approx - 60 %)$ for heavy metals. However, $β$ - $W$ has a limitation when one is aiming for reliable technology integration: the $β$ phase is limited to a thickness of a few nanometers and enters the $α$ phase above 4 nm in our samples when industry-relevant deposition tools are used. Here, we report our approach to extending the range of $β$ - $W$ , while simultaneously improving the SOT efficiency by introducing $N$ and $O$ doping of $W$ . Resistivity and XRD measurements confirm the extension of the $β$ phase from 4 nm to more than 10 nm, and transport characterization shows an effective SOT efficiency larger than $- 44.4 %$ (reaching approximately $- 60 %$ for the bulk contribution). In addition, we demonstrate the possibility of controlling and enhancing the perpendicular magnetic anisotropy of a storage layer ( $Co − Fe − B$ ). Further, we integrate the optimized $W (O, N)$ into SOT magnetic random-access memory (SOT-MRAM) devices and project that, for the same thickness of SOT material, the switching current decreases by 25% in optimized $W (O, N)$ compared with our standard $W$ . Our results open the path to using and further optimizing $W$ for integration of SOT-MRAM technology.

Received 16 September 2021
Accepted 4 November 2021
Corrected 20 December 2021

DOI:https://doi.org/10.1103/PhysRevApplied.16.064009

Physics Subject Headings (PhySH)

Spin torque Spin-orbit coupling Spintronics

Magnetic tunnel junctions

Condensed Matter, Materials & Applied Physics

Corrections

20 December 2021

Correction: The fourth affiliation contained an error and has been set right.

Authors & Affiliations

Kiran Kumar Vudya Sethu^1,2,*, Sambit Ghosh¹, Sebastien Couet¹, Johan Swerts¹, Bart Sorée^1,2,3, Jo De Boeck^1,2, Gouri Sankar Kar¹, and Kevin Garello ^1,4,†

¹imec, Kapeldreef 75, Leuven 3001, Belgium
²KU Leuven, Department of Electrical Engineering, Kasteelpark Arenberg 10, Leuven 3001, Belgium
³University of Antwerp, Physics Department, Groenenborgerlaan 171, Antwerpen B-2020, Belgium
⁴Université Grenoble Alpes, CEA, CNRS, Grenoble INP, SPINTEC, 38054 Grenoble, France

^*kiran.kumar.vudyasethu@imec.be
^†kevin.garello@cea.fr

Article Text (Subscription Required)

Click to Expand

References (Subscription Required)

Click to Expand

Issue

Vol. 16, Iss. 6 — December 2021

Subject Areas

Reuse & Permissions

Access Options

Author publication services for translation and copyediting assistance advertisement

Images

Figure 1
(a) Sample stacks prepared for this study. The main difference between the samples is in the tungsten layer. Sample 1 (S1) is deposited with nondoped $W$ . In sample 2 (S2), the $W$ layer is doped with oxygen. This is denoted as $W (O)$ . This layer is grown as nanolaminate repeats. The $W$ layer in sample 3 (S3) is grown by reactive doping with nitrogen. This is denoted as $W (N)$ . In sample 4 (S4), the $W$ is prepared by doping with both nitrogen and oxygen. This is denoted as $W (O, N)$ . The $Co − Fe − B$ is capped with $Mg O$ (1 nm), $Co − Fe − B$ dusting, and $Ta$ capping layers. Slanted layers indicate layers deposited as a wedge on a single wafer. (b) Schematic illustration showing Hall-bar measurement geometry along with definition of the coordinates used in the harmonic Hall-voltage measurements for the quantification of SOTs in our samples. (c) Schematic illustration of electrical measurements of SOT-MRAM cell. $V_{SOT}$ is a voltage pulse with a pulse width $t_{p, SOT}$ , applied through the SOT track for electrical switching of the free layer of the MTJ. The orientation of the MTJ (parallel or antiparallel) is measured by applying a dc voltage through the MTJ. dc TMR measurements are performed to measure the hysteresis loop of the free layer. SAF, synthetic anti-ferromagnet; RL, reference layer.
Reuse & Permissions
Figure 2
(a) Sample resistivity $ρ$ estimated from four-point-probe sheet-resistance measurements as a function of $W$ layer thickness. (b) Grazing-incidence XRD spectra of samples. The thickness of the $W$ layer, along with the sample type, is indicated in the legend. Peak positions corresponding to different crystal phases of tungsten are indicated, with reference to the lattice plane.
Reuse & Permissions
Figure 3
(a) Saturation magnetization of $Co − Fe − B$ as a function of tungsten layer thickness. Sample types are indicated in the legend. (b) Effective anisotropy field, $B_{K}^{eff}$ , of $Co − Fe − B$ vs $W$ thickness measured at Hall-bar level as described in the text. Positive values of $B_{K}^{eff}$ indicate PMA, and negative values correspond to a demagnetization field and therefore IMA.
Reuse & Permissions
Figure 4
(a) Antidamping SOT efficiency, $ξ_{J}^{AD}$ , of different types of $W$ as a function of tungsten layer thickness. The solid lines are fits to the experimental data points using a drift-diffusion model to extract the intrinsic spin Hall angle $θ_{SH}^{eff}$ and the spin diffusion length $λ_{sf}$ . (b) Fieldlike SOT efficiency, $ξ_{J}^{FL}$ , as a function of SOT-layer thickness $t_{SOT}$ .
Reuse & Permissions
Figure 5
$ξ_{J_{SOT}}^{AD}$ renormalized to the current flowing through the SOT layer, $I_{SOT}$ . The solid lines are a fit of a drift-diffusion model to the data.
Reuse & Permissions
Figure 6
(a) Hysteresis loops of the free layer of a PMTJ SOT-MRAM cell with $W$ and $W (O, N)$ as SOT layers. (b) Electrical-switching (SOT) probability distribution for a 60-nm MTJ with $W$ and $W (O, N)$ as SOT-generating layers, as a function of SOT current $I$ . The width of the SOT pulse used is 1 ns. An in-plane field ( $B_{X}$ ) of 32 mT is applied during the switching measurements. (c) Pulse-width dependence of threshold current $I_{C}$ for switching a 60-nm PMTJ by SOT, for $W$ and $W (O, N)$ SOT layers. (d) Dependence of $I_{C 0}$ on in-plane magnetic field $| B_{X} |$ for $W$ and $W (O, N)$ SOT layers in a PMTJ SOT-MRAM cell.
Reuse & Permissions

Physical Review Applied

Optimization of Tungsten β-Phase Window for Spin-Orbit-Torque Magnetic Random-Access Memory

Kiran Kumar Vudya Sethu, Sambit Ghosh, Sebastien Couet, Johan Swerts, Bart Sorée, Jo De Boeck, Gouri Sankar Kar, and Kevin Garello

Phys. Rev. Applied 16, 064009 – Published 3 December 2021

Abstract

Physics Subject Headings (PhySH)

Corrections

Authors & Affiliations

Article Text (Subscription Required)

References (Subscription Required)

Issue

Subject Areas

Access Options

Authorization Required

Other Options

Download & Share

Images

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Log In

Search

Article Lookup

Paste a citation or DOI

Enter a citation

Optimization of Tungsten $β$ -Phase Window for Spin-Orbit-Torque Magnetic Random-Access Memory