Robin B. Jacobs-Gedrim

Analog in-memory computing (AIMC) utilizing nonvolatile resistive memory arrays can drastically reduce the energy required for neural network inference. The oxide-based resistive random access memory (ReRAM) devices are a promising candidate for AIMC due to their high density and ability to store multiple bits of precision in an analog state. However, variability in conductance fluctuations across devices and programming cycles in each device presents a challenge for inference accuracy. We characterize the signature of conductance fluctuations in a Ta/Ta₂O_5-x ReRAM array, finding significant variability in fluctuations from cycle-to-cycle associated with different filament configurations, which ultimately causes high conductance variability at the conclusion of the settling period of $\approx 1$ h following programming. Although cycle-to-cycle variability and device-to-device fluctuation variability were similar, specific devices tended toward high cycle-to-cycle variability. Most devices had an optimal filamentary configuration, which was relatively stable over time. By creating a smart write-verify routine, which determines the optimal filament configuration, it was possible to improve the inference accuracy for CIFAR-10 image recognition from 28.5% to 79.1% as measured following the 1-h settling period.

We study the impact of irradiation on magnetic tunnel junction (MTJ) films with perpendicular magnetic anisotropy (PMA) and spin-orbit torque (SOT) switching using magneto-optical Kerr effect and transmission electron microscopy. Our results show that the thin-film stack is robust to gamma ionizing dose up to 1 Mrad(Si) and Ta ¹⁺ ion irradiation fluences up to 10 ¹² ions/cm ² , showing SOT PMA MTJs are radiation-hard. But, at very high Ta ¹⁺ ion irradiation between 10 ¹² and 10 ¹⁴ ions/cm ² , reduced coercivity and eventually greatly reduced PMA are observed, corresponding with an increase in intermixing of the CoFeB-MgO layers, particularly at the lower CoFeB-MgO interface. These results agree with displacement damage modeling that predicts higher damage in the layers closer to the bottom heavy metal and substrate. Compared to spin transfer torque and in-plane anisotropy MTJs, needing a top-pinned stack, a thicker heavy metal layer, and perpendicular anisotropy that is pinned out of plane by interfaces, all could make SOT PMA MTJs more susceptible to damage at high doses.

We evaluate the resilience of CoFeB/MgO/CoFeB magnetic tunnel junctions (MTJs) with perpendicular magnetic anisotropy (PMA) to displacement damage induced by heavy-ion irradiation. MTJs were exposed to 3-MeV Ta ²⁺ ions at different levels of ion beam fluence spanning five orders of magnitude. The devices remained insensitive to beam fluences up to 10 ¹¹ ions/cm ² , beyond which a gradual degradation in the device magnetoresistance, coercive magnetic field, and spin-transfer-torque (STT) switching voltage were observed, ending with a complete loss of magnetoresistance at very high levels of displacement damage (>0.035 displacements per atom). The loss of magnetoresistance is attributed to structural damage at the MgO interfaces, which allows electrons to scatter among the propagating modes within the tunnel barrier and reduces the net spin polarization. Ion-induced damage to the interface also reduces the PMA. This study clarifies the displacement damage thresholds that lead to significant irreversible changes in the characteristics of STT magnetic random access memory (STT-MRAM) and elucidates the physical mechanisms underlying the deterioration in device properties.

The domain-wall (DW)-magnetic tunnel junction (MTJ) device implements universal Boolean logic in a manner that is naturally compact and cascadable. However, an evaluation of the energy efficiency of this emerging technology for standard logic applications is still lacking. In this article, we use a previously developed compact model to construct and benchmark a 32-bit adder entirely from DW-MTJ devices that communicates with DW-MTJ registers. The results of this large-scale design and simulation indicate that while the energy cost of systems driven by spin-transfer torque (STT) DW motion is significantly higher than previously predicted, the same concept using spin-orbit torque (SOT) switching benefits from an improvement in the energy per operation by multiple orders of magnitude, attaining competitive energy values relative to a comparable CMOS subprocessor component. This result clarifies the path toward practical implementations of an all-magnetic processor system.

The radiation response of TaOx-based resistive memory (RRAM) devices fabricated in academic (Set A) and industrial (Set B) settings was compared. Ionization damage from a ⁶⁰ Co gamma source did not cause any changes in device resistance for either device type, up to 45 Mrad(Si). Displacement damage from a heavy ion beam caused a decrease in resistance at 1× 10 ²¹ oxygen displacements per cm ³ in Set B devices in the high-resistance state (HRS); meanwhile, Set A devices did not exhibit any decrease in resistance due to displacement damage. Both types of devices exhibited an increase in resistance around 3 × 10 ²² oxygen displacements per cm ³ , possibly due to the damage at the oxide/metal interfaces. These extremely high levels of damage represent near-total atomic disruption, and if this level of damage was ever reached, other circuit elements would likely fail before the RRAM devices in this article. Overall, both sets of devices were much more resistant to radiation effects than the similar devices reported in the literature. Displacement damage effects were only observed in the Set A devices once the displacement-induced oxygen vacancies surpassed the intrinsic vacancy concentration in the devices, suggesting that high oxygen vacancy concentration played a role in the devices' high tolerance to displacement damage.

The image classification accuracy of a TaOx ReRAM based neuromorphic computing accelerator is evaluated after intentionally inducing displacement damage up to a fluence of 1014 2.5 MeV Si ions / cm2 on the analog devices used to store weights. Results are consistent with a radiation induced oxygen vacancy production mechanism. When the device is in the High Resistance State (HRS) during heavy ion radiation, the device resistance, linearity, and accuracy after training are only affected by high fluence levels. The findings in this study are in accordance with results of previous studies on TaOx based digital resistive random-access memory. When the device is in the Low Resistance State (LRS) during irradiation, no resistance change was detected, but devices with a 4 kΩ in line resistor did show a reduction in accuracy after training at 1014 2.5 MeV Si ions / cm2. This indicates that changes in resistance can only be somewhat correlated with changes to a devices’ analog properties. This study demonstrates that TaOx devices are radiation tolerant not only for high radiation environment digital memory applications, but also when operated in an analog mode suitable for neuromorphic computation and training on new datasets.

Analog nonvolatile resistive switching phenomena in metal oxides can potentially be used as a synaptic weight in hardware based neuromorphic computing accelerators. Single layer resistive random-access memory (RRAM) devices have switching currents in the greater than 1 mA range, effectively requiring too much energy for integration in a crossbar array based neural accelerator. This study details the fabrication and characterization of a bilayer RRAM device consisting of a Pt-TaO_x-Al₂O₃-TiN stack which is designed for low current operation. This high resistance bilayer device reduces switching energy to ∼8 pJ during RESET and 15 pJ during SET, at the expense of increased operational noise. Noise increase is expected in this higher resistance device due to electron trapping in levels created by vacancies piling up at the interface between the Al₂O₃ and TaO_x layer. As a result, the simulated performance of these devices used in training a neuromorphic accelerator on the MNIST dataset was 80%, significantly lower than required. Using the difference in current between two devices to represent a digit and using two digits per weight with a technique called periodic carry (for a total of 4 devices), a training accuracy of 93% could be achieved. The device and methods detailed here represent a necessary step toward the realization of energy efficient neuromorphic accelerators.