Reduced sensitivity to process, voltage and temperature variations in activated perpendicular magnetic tunnel junctions based stochastic devices

Md Golam Morshed,Laura Rehm,Ankit Shukla,Yunkun Xie,Samiran Ganguly,Shaloo Rakheja,Andrew D. Kent,Avik W. Ghosh

2023-10-29

Abstract:True random number generators (TRNGs) are fundamental building blocks for many applications, such as cryptography, Monte Carlo simulations, neuromorphic computing, and probabilistic computing. While perpendicular magnetic tunnel junctions (pMTJs) based on low-barrier magnets (LBMs) are natural sources of TRNGs, they tend to suffer from device-to-device variability, low speed, and temperature sensitivity. Instead, medium-barrier magnets (MBMs) operated with nanosecond pulses - denoted, stochastic magnetic actuated random transducer (SMART) devices - are potentially superior candidates for such applications. We present a systematic analysis of spin-torque-driven switching of MBM-based pMTJs (Eb ~ 20 - 40 kBT) as a function of pulse duration (1 ps to 1 ms), by numerically solving their macrospin dynamics using a 1-D Fokker-Planck equation. We investigate the impact of voltage, temperature, and process variations (MTJ dimensions and material parameters) on the switching probability of the device. Our findings indicate SMART devices activated by short-duration pulses (< 1 ns) are much less sensitive to process-voltage-temperature (PVT) variations while consuming lower energy (~ fJ) than the same devices operated with longer pulses. Our results show a path toward building fast, energy-efficient, and robust TRNG hardware units for solving optimization problems.

Applied Physics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: how to reduce the sensitivity of perpendicular magnetic tunnel junctions (pMTJs) to process, voltage and temperature variations in the application of True Random Number Generators (TRNGs). Specifically, the paper focuses on stochastic devices based on medium - barrier magnets (MBMs) and explores the performance of these devices under different conditions.

### Main problems and background

1. **Importance of TRNGs**:
- TRNGs are widely used in fields such as cryptography, Monte Carlo simulation, neuromorphic computing and probabilistic computing.
- The pseudo - random number sequences generated by traditional software algorithms and CMOS technology have correlations and predictability and cannot meet the requirements of TRNGs.

2. **Limitations of existing technologies**:
- **High - barrier magnets (HBMs)**: Although stable, they have high energy consumption and low throughput.
- **Low - barrier magnets (LBMs)**: They are energy - efficient, but have a slow dynamic response, are sensitive to process and temperature variations, and are difficult to manufacture in practice.

3. **Advantages of MBM**:
- MBMs combine the advantages of HBMs and LBMs, have lower energy consumption and higher reliability, and are suitable for constructing fast, efficient and robust TRNG hardware units.

### Research content

The paper systematically analyzes the spin - torque - driven switching behavior of MBM - based pMTJs under different pulse durations (from 1 ps to 1 ms) by numerically solving the one - dimensional Fokker - Planck equation, and studies the influence of voltage, temperature and process variations (such as MTJ size and material parameters) on the switching probability.

### Key findings

1. **SMART devices activated by short pulses**:
- Under short pulses (<1 ns), SMART devices have lower sensitivity to process, voltage and temperature variations and lower energy consumption (about several femto - joules).

2. **Energy efficiency and robustness**:
- SMART devices activated by short pulses can maintain high robustness while achieving low energy consumption, providing a possible path for constructing fast, efficient and reliable TRNG hardware units.

### Formula display

- **Landau - Lifshitz - Gilbert (LLG) equation**:
\[
\frac{1+\alpha^{2}}{\gamma}\cdot\frac{\partial m}{\partial t}=-\mu_{0}\cdot(m\times H_{\text{eff}})-\alpha\mu_{0}\cdot m\times(m\times H_{\text{eff}})
\]
\[
-\frac{\hbar}{2e}\cdot\eta I\cdot\frac{m\times(m\times m_{p})}{M_{s}\Omega}
\]

- **Fokker - Planck (FP) equation**:
\[
\frac{\partial\rho}{\partial t}=-\nabla\cdot(L\rho)+D_{i}\nabla^{2}\rho
\]
where \(\rho(\theta; t)\) is the probability density of the magnetization intensity at time \(t\), \(\theta\) is the magnetization angle, \(L\) is the sum of all effective moments, \(D_{i}\) is the effective diffusion constant considering thermal fluctuations, defined as:
\[
D_{i}=\frac{\alpha\gamma k_{B}T}{(1 + \alpha^{2})\mu_{0}M_{s}\Omega}
\]

- **Switching probability formula**:
\[
P_{\text{sw}}=