TL;DR
This paper investigates how the "tunneling magnetoresistance" (TMR) effect in magnetic tunnel junctions (MTJs) changes with temperature. They found that annealing the MTJs significantly improves TMR and alters its temperature dependence, suggesting a modified Julliere model with spin-dependent two-step tunneling is needed to explain these observations.
What Problem Does This Paper Address?
Magnetic tunnel junctions (MTJs) are crucial components for novel electronic devices like magnetic random-access memory (MRAM). The tunneling magnetoresistance (TMR) effect, where the electrical resistance changes based on the relative magnetic alignment of two ferromagnetic layers separated by a thin insulator, is the core phenomenon behind these devices. While TMR is well-established, understanding its behavior across different temperatures is vital for designing reliable and high-performance devices. Previous studies often showed TMR decreasing with increasing temperature. This paper addresses the question of why, in their specific MTJ structure, TMR increases with temperature before annealing, and how annealing affects this behavior. They aim to explain this non-monotonic temperature dependence, which deviates from simpler models.
Methods
The researchers fabricated MTJs using a sputtering system. The MTJ structure consisted of several layers, including a CoFe/Al₂O₃/CoFe core for the tunnel barrier, sandwiched between ferromagnetic layers and buffer/capping layers. The Al₂O₃ barrier was formed by oxidizing a thin Al layer. Crucially, they prepared samples in two states: "as-grown" and "annealed" at 200°C. They measured the TMR by applying a magnetic field and using a DC four-probe method with a low DC bias (1 mV). Temperature dependence was studied by measuring TMR and conductance at various temperatures. To explain their findings, they proposed a modified Julliere model incorporating a "spin-dependent two-step tunneling" mechanism.
Datasets
Not discussed in the paper. The study focuses on experimental measurements from fabricated MTJ samples.
Key Insights and Ideas
- Annealing is a game-changer: Annealing the MTJs significantly boosted the TMR ratio (from 20.7% to 59.0%) and improved the overall shape of the MR curve. This suggests annealing heals defects or modifies the barrier, leading to better performance.
- Unusual pre-annealing TMR behavior: Before annealing, the TMR increased with temperature up to 160 K, a behavior contrary to typical observations where TMR decreases with temperature. This points to complex underlying physics.
- Conductance hints at defects: The significant increase in junction conductance with temperature before annealing suggests the presence of impurity or defect states within the Al₂O₃ barrier, which are more sensitive to thermal distribution.
- Spin-dependent two-step tunneling is key: The authors propose that a modified Julliere model, accounting for spin-dependent two-step tunneling through these defect states, is necessary to explain the observed temperature dependence, especially the pre-annealing increase in TMR.
- Post-annealing behavior is more conventional: After annealing, the TMR ratio decreased with temperature, aligning with more commonly reported trends, indicating that annealing likely reduced the influence of problematic defect states.
Key Results
- TMR Ratio: Increased from 20.7% to 59.0% after annealing at 100 K.
- Temperature Dependence (Pre-annealing): TMR increased with temperature from 80 K to 160 K.
- Temperature Dependence (Post-annealing): TMR decreased with temperature, consistent with typical observations.
- Conductance: Increased significantly (about 80%) with temperature (80 K to 300 K) for both parallel (P) and antiparallel (AP) alignments before annealing. Annealing also increased junction conductance.
Limitations and Future Work
- Model Complexity: The proposed "spin-dependent two-step model" is a phenomenological explanation. Further theoretical work is needed to fully elucidate the microscopic origins of the spin-dependent transfer rates and defect states.
- Defect Characterization: While defects are implicated, their precise nature and location within the barrier are not fully characterized. Future work could involve more detailed material analysis.
- Annealing Optimization: The study used a single annealing condition (200°C for 1 hour). Optimizing annealing parameters could potentially lead to even higher TMR values or further refine the temperature dependence.
- Device Scaling: The paper mentions junction sizes of 50 µm × 50 µm. Future research could explore how these temperature effects scale down to the smaller dimensions required for practical MRAM applications.
Summary updated