Dimensional Scaling of Ferroelectric Properties of Hafnia-Zirconia Thin Films: Electrode Interface Effects.

Hafnia-based ferroelectric (FE) thin films are promising candidates for semiconductor memories. However, a fundamental challenge that persists is the lack of understanding regarding dimensional scaling, including thickness scaling and area scaling, of the functional properties and their heterogeneity in these films. In this work, excellent ferroelectricity and switching endurance are demonstrated in 4 nm-thick Hf0.5Zr0.5O2 (HZO) capacitors with molybdenum electrodes in capacitors as small as 65 nm × 45 nm in size. The HZO layer in these capacitors can be crystallized into the ferroelectric orthorhombic phase at the low temperature of 400 °C, making them compatible for back-end-of-line (BEOL) FE memories. With the benefits of thickness scaling, low operation voltage (1.2 V) is achieved with high endurance (>1010 cycles); however, a significant fatigue regime is noted. We observed that the bottom electrode, rather than the top electrode, plays a dominant role in the thickness scaling of HZO ferroelectric behavior. Furthermore, ultrahigh switched polarization (remanent polarization 2Pr ∼ 108 µC cm-2) is observed in some nanoscale devices. This study advances the understanding of dimensional scaling effects in HZO capacitors for high-performance FE memories.

Area-Selective Atomic Layer Deposition for Resistive Random-Access Memory Devices.

Resistive random-access memory (RRAM) is a promising technology for data storage and neuromorphic computing; however, cycle-to-cycle and device-to-device variability limits its widespread adoption and high-volume manufacturability. Improving the structural accuracy of RRAM devices during fabrication can reduce these variabilities by minimizing the filamentary randomness within a device. Here, we studied area-selective atomic layer deposition (AS-ALD) of the HfO2 dielectric for the fabrication of RRAM devices with higher reliability and accuracy. Without requiring photolithography, first we demonstrated ALD of HfO2 patterns uniformly and selectively on Pt bottom electrodes for RRAM but not on the underlying SiO2/Si substrate. RRAM devices fabricated using AS-ALD showed significantly narrower operating voltage range (2.6 × improvement) and resistance states than control devices without AS-ALD, improving the overall reliability of RRAM. Irrespective of device size (1 × 1, 2 × 2, and 5 × 5 µm2), we observed similar improvement, which is an inherent outcome of the AS-ALD technique. Our demonstration of AS-ALD for improved RRAM devices could further encourage the adoption of such techniques for other data storage technologies, including phase-change, magnetic, and ferroelectric RAM.

Atomistic study of dynamics for metallic filament growth in conductive-bridge random access memory.

The growth dynamics for metallic filaments in conductive-bridge resistive-switching random access memory (CBRAM) are studied using the kinetic Monte Carlo (KMC) method. The physical process at the atomistic level is revealed in explaining the experimental observation that filament growth can originate at either the cathode or the anode. The statistical nature of the filament growth is best shown by the random topography of dendrite-like conductive paths obtained. Critical material properties, such as charged-particle mobility in the switching layer of a solid electrolyte or a dielectric, are mapped to KMC model parameters through activation energy, etc. The accuracy of the simulator is established by the good agreement between the simulated forming time and the measured data.

A physics/circuit-based switching model for carbon-based resistive memory with sp2/sp3 cluster conversion.

The switching mechanism in carbon-based resistive-switching random access memory is modelled using a percolation approach built on the low-temperature transition between phases sp(3) (diamond-like and high-resistive state) and sp(2) (graphite-like and low-resistive state) for a matrix of carbon clusters in a diamond-like carbon film. The switching process is described using a random circuit breaker network with each breaker controlled by the resistance of clusters sp(2)/sp(3). The key feature of the proposed model is the thermal stress-induced transition from sp(2) to sp(3) phase and the electric field-induced transition from sp(3) to sp(2) phase. Compared with experiments on the switching biasing scheme, a good agreement between simulation and measured data validated the accuracy of the proposed model.