Voltage-insensitive stochastic magnetic tunnel junctions with double free layers

Rikuto Ota,Keito Kobayashi,Keisuke Hayakawa,Shun Kanai,Kerem Y. Çamsarı,Hideo Ohno,Shunsuke Fukami

2024-05-31

Abstract:Stochastic magnetic tunnel junctions (s-MTJ) is a promising component of probabilistic bit (p-bit), which plays a pivotal role in probabilistic computers. For a standard cell structure of the p-bit, s-MTJ is desired to be insensitive to voltage across the junction over several hundred millivolts. In conventional s-MTJs with a reference layer having a fixed magnetization direction, however, the stochastic output significantly varies with the voltage due to spin-transfer torque (STT) acting on the stochastic free layer. In this work, we study a s-MTJ with a "double-free-layer" design theoretically proposed earlier, in which the fixed reference layer of the conventional structure is replaced by another stochastic free layer, effectively mitigating the influence of STT on the stochastic output. We show that the key device property characterized by the ratio of relaxation times between the high- and low-resistance states is one to two orders of magnitude less sensitive to bias voltage variations compared to conventional s-MTJs when the top and bottom free layers are designed to possess the same effective thickness. This work opens a pathway for reliable, nanosecond-operation, high-output, and scalable spintronics-based p-bits.

Mesoscale and Nanoscale Physics

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is: **How to design a voltage - insensitive stochastic magnetic tunnel junction (s - MTJ) to improve its reliability and scalability in probabilistic computing?**

Specifically, when a voltage is applied to a traditional single - free - layer (SFL) stochastic magnetic tunnel junction, its random output will change significantly due to the influence of spin - transfer torque (STT). This voltage sensitivity limits its application in probabilistic bits (p - bits), especially in probabilistic computers that require high - precision control of the output.

To solve this problem, this paper studies a stochastic magnetic tunnel junction with a "double - free - layer" (DFL) structure. In this structure, the fixed reference layer in the traditional structure is replaced by another random free layer, thereby effectively reducing the influence of STT on the random output. Experimental results show that when the effective thicknesses of the two free layers are similar, the relaxation - time ratio (\(\frac{\tau_{AP}}{\tau_P}\)) of DFL s - MTJ shows extremely low sensitivity to the change of bias voltage. Compared with the traditional SFL s - MTJ, its voltage sensitivity is reduced by 1 to 2 orders of magnitude.

In this way, DFL s - MTJ can achieve more reliable nanosecond - level operations, high output, and scalable spintronics - based probabilistic bits (p - bits), providing a new approach for future probabilistic computing hardware.

### Key Formulas

1. **Néel - Arrhenius Law**:
\[
\tau_{P/AP}=\tau_0\exp\left[\Delta_{P/AP}\right]
\]
where,
\[
\Delta_{P/AP}=\frac{M_sH_{k,in}v}{2k_BT}\left(1 + \frac{H_x - H_s}{H_{k,in}}\right)^n\left(1-\frac{V}{V_{c0,P/AP}}\right)^m
\]

2. **Relaxation - Time Ratio of Double - Free - Layer Structure**:
\[
\frac{\tau_{AP}}{\tau_P}=\frac{\tau_{1 +,2 -}+\tau_{1 -,2 +}}{\tau_{1 +,2 +}+\tau_{1 -,2 -}}
\]
When the two free layers are equivalent:
\[
\frac{\tau_{AP}}{\tau_P}=1 + O(V^2)
\]

These formulas show that by precisely controlling the thicknesses of the two free layers, the random behavior of DFL s - MTJ can be made insensitive to voltage changes, thereby improving its performance and reliability in probabilistic computing.