Editor’s note: This work was first presented at the 2011 IEEE International Electron Devices Meeting (IEDM) and appears here courtesy of the IEEE. For more information about IEDM 2012 (San Francisco, CA; December 10-12), click here.
In this article, we report the first experimental demonstration of sub-20-nm magnetic tunnel junction (MTJ) cells for investigating the downscaling feasibility of spin-transfer torque (STT) MRAM, one of the most promising candidates to replace conventional memories. We demonstrate the STT switching of 17-nm node P-MTJ cells, the smallest feature size ever reported, utilizing perpendicular materials possessing high interface anisotropy of 2.5 erg/cm2
and improved integration processes to achieve reproducible switching with critical current (Ic) of 44 uA, tunneling magneto-resistance (TMR) ratio of 70% and thermal stability factor (E/kB
T) of 34.
One of the critical tasks in developing scalable STT- MRAM for future standalone memory lies in validating the STT switching characteristics sustainable at sub-20-nm node where the conventional memories are expected to meet with serious scaling problems. It is widely recognized that perpendicular magnetic anisotropy (PMA) should be implemented into MTJs at such a small node to obtain high enough thermal stability. Although many PMA materials have been studied, for instance L10
ordered alloys (FePt and CoPt, etc), Co-based multilayers (Co/Pd and Co/Pt, etc) and rare-earth and transition metal alloys [1, 2], utilizing the PMA materials as a free layer in the MTJs is severely challenging . We demonstrate the feasibility of a sub-20-nm MRAM device adopting the free layer of interface driven PMA (i- PMA) [4, 5] with enhanced anisotropy energy without using additional PMA materials.MTJ design through PMA enhancement
The effective materials system for enhancing the anisotropy energy at the i-PMA scheme can be predicted by first-principle calculations . For estimating the effects of oxidation process, we calculated anisotropy energy constants (Ki = Ku×t) for different levels of oxidation by the first-principle method using a VASP code. Figure 1 shows the calculated Ki values for various amounts of oxygen at the interface of FeCo and MgO. The result reveals that the degree of Mg oxidation at the interface is a crucial parameter to determine the anisotropy energy. Figure 1: (a) Simulated structures and (b) calculated values of interface-driven perpendicular magnetic anisotropy (Ki.calc.) as a function of oxygen amount at the interface between FeCo and MgO.
For experimental realization of i-PMA, two specialized methods of forming MgO have been evaluated. The half-MTJ stack of Ta(5-nm)/Co20
(t)/MgO/Ta(5-nm) on thermally oxidized Si(001) was used. Figure 2 compares the Ki values as a function of CoFeB thickness for the two different MgO formation processes. It reveals that process-(1) and (2) produce PMA under the different critical thickness of 1.3 and 1.1 nm, respectively. The maximum Ki values at the peak position extracted from the two processes are 0.54 and 0.14 erg/cm2
and the extrapolated Ki values at y-intercept for the two processes are 2.5 and 1.3 erg/cm2
Figure 2: Measured values of interface-driven perpendicular magnetic anisotropy value (Ki. Meas) as a function of CoFeB thickness deposited on process-1 MgO and process-2 MgO.
Figure 3 shows the minimum required values of the effective Ki as a function of MTJ cell size with aspect ratio of 1 and 2, under single domain assumption, for making the thermal stability factor (Δ=E/kB
T) higher than 70. In this figure, it is recognized that for downscaling below 20-nm MTJ cell with aspect ratio of 2, Ki value should be higher than 0.46 erg/cm2
. The standard MTJ samples with the structure of Ta(5)/Co20
/MgO/pinned layer were used to measure TMR ratio (%) and resistance-area (RA) product using a current in-plane tunneling method (CIPT). The standard MTJ samples with the structure of Ta(5)/Co20
B20/MgO/pinned layer were used to measure TMR ratio (%) and resistance-area (RA) product using a current in-plane tunneling method (CIPT).
Figure 3: Required values of i-PMA energy constants for satisfying high thermal stability factors (Δ = Ku*V/kB*T > 70) as a function of MTJ cell size with aspect ratios 1 and 2.
Figure 4: TMR ratio and i-PMA energy constants as a function of CoFeB thickness, measured at a standard MTJ sample.
Figure 5 shows that the relationship between TMR ratio and RA product as a function of oxidation process. At the point of just-oxidation, maximum TMR and minimum RA values are achieved.
Figure 5: TMR ratios and RA products measured at standard MTJ samples by varying oxidation time.
In-plane saturation field Hk and coercivity Hc of i-PMA samples are shown in figure 6 (a) and (b), respectively. Based on the results, i-PMA MTJ structures with sub-20-nm scalability have been designed adopting the layers of Δ~70, Ki~0.53 erg/cm2
, TMR ratio ~100% and RA product ~ 8 Ω
.Figure 6: (a) In-plane saturation field, Hk, (b) coercivity, Hc at various CoFeB thickness.