The utilization of bipolar-type memristive devices for the realization of synaptic connectivity in neural networks strongly depends on the ability of the devices for analog conductance modulation under application of electrical stimuli in the form of identical voltage pulses. Typically, filamentary valence change mechanism (VCM)-type devices show an abrupt SET and a gradual RESET switching behavior. Thus, it is challenging to achieve an analog conductance modulation during SET and RESET. Here, we show that analog as well as binary conductance modulation can be achieved in a Pt/HfO2/TiOx/Ti VCM cell by varying the operation conditions. By analyzing the switching dynamics over many orders of magnitude and comparing to a fully dynamic switching model, the origin of the two different switching modes is revealed. SET and RESET transition show a two-step switching process: a fast conductance change succeeds a slow conductance change. While the time for the fast conductance change, the transition time, turns out to be state-independent for a specific voltage, the time for the slow conductance change, the delay time, is highly state-dependent. Analog switching can be achieved if the pulse time is a fraction of the transition time. If the pulse time is larger than the transition time, the switching becomes probabilistic and binary. Considering the effect of the device state on the delay time in addition, a procedure is proposed to find the ideal operation conditions for analog switching.

Redox-type resistive random access memories based on transition-metal oxides are studied as adjustable two-terminal devices for integrated network applications beyond von Neumann computing. The prevailing, so-called, counter-eight-wise (c8w) polarity of the switching hysteresis in filamentary-type valence change mechanism devices originates from a temperature- and field-controlled drift-diffusion process of mobile ions, predominantly oxygen vacancies in the switching oxide. Recently, a bipolar resistive switching (BRS) process with opposite polarity, so-called, eight-wise (8w) switching, has been reported that, especially for TiO cells, is still not completely understood. Here, we report on nanosized (<0.01 μm) asymmetric memristive cells from 3 to 6 nm thick TiO films by atomic layer deposition, which reveal a coexistence of c8w and 8w switching in the same cell. As important characteristics for the studied Pt/TiO/Ti/Pt devices, the resistance states of both modes are nonvolatile and share one common state; i.e., the high-resistance state of the c8w mode equals the low-resistance state of the 8w-mode. A transition between the opposite hysteresis loops is possible by voltage control. Specifically, 8w BRS in the TiO cells is a self-limited low-energy nonvolatile switching process. Additionally, the 8w reset process enables the programming of multilevel high-resistance states. Combining the experimental results with data from simulation studies allows to propose a model, which explains 8w BRS by an oxygen transfer process across the Pt/TiO Schottky interface at the position of the c8w filament. Therefore, the coexistence of c8w and 8w BRS in the nanoscale asymmetric Pt/TiO/Ti/Pt cells is understood from a competition between drift/diffusion of oxygen vacancies in the oxide layer and an oxygen exchange reaction across the Pt/TiO interface.

No abstract

Future development of the modern nanoelectronics and its flagships internet of things, artificial intelligence, and neuromorphic computing is largely associated with memristive elements, offering a spectrum of inevitable functionalities, atomic level scalability, and low-power operation. However, their development is limited by significant variability and still phenomenologically orientated materials’ design strategy. Here, we highlight the vital importance of materials’ purity, demonstrating that even parts-per-million foreign elements substantially change performance. Appropriate choice of chemistry and amount of doping element selectively enhances the desired functionality. Dopant/impurity-dependent structure and charge/potential distribution in the space-charge layers and cell capacitance determine the device kinetics and functions. The relation between chemical composition/purity and switching/neuromorphic performance is experimentally evidenced, providing directions for a rational design of future memristive devices.